01
Understanding to ImproveG E T T I N G T H E M O S T O U T O F C U S T O M E R amp E M P L O Y E E
D A T A W I T H W O R L D - C L A S S T E X T A N A L Y T I C S
02
IntroductionText analytics are vital to your brandrsquos ability to understand your
customer and employee experiences You can have listening posts
across every channel and at every point in the customer journey
but if you donrsquot have the best possible text analytics solution in
place your ability to derive actionable intelligence from that data
is essentially moot And your ability to create transformational
change across the organization and drive business growth Thatrsquod
be a non-starter without effective text analytics Without them
all you have is a score not any context or information on what
actually went well or needs improvement
Itrsquos obvious that text analytics are vital but in an industry full of
jargon claims about accuracy and a huge amount of conflicting
data how can you tell what solution attributes will be the best for
your company
Well in this eBook wersquore cutting through the fancy lingo and
salesy claims to get to the heart of what makes a text analytics
solution truly successful Keep reading to learn more about top
terms accuracy solution necessities and more
Without text analytics all you have is a score not any context or information on what actually went well or needs improvement
03
TABL
E O
F CO
NTE
NTS Pg 04 Pg 05 Pg 09
Pg 21Pg 17Pg 13
Pg 24
TEXT ANALYTICS 101
A QUICK GLOSSARY
A ldquoMODELrdquo TEXT
ANALYTICS SOLUTIONFACTORING IN SENTIMENTGETTING GLOBAL
WITH TRANSLATION
DRIVING IMPROVEMENT
LETrsquoS TALK ABOUT
ACCURACY
TOP OBJECTIVES FOR A
MODERN TA SOLUTION
04TEXT ANALY TICS 101A Quick Glossary
Therersquos no way around it conversations around text analytics are chock-full of jargon Below yoursquoll find straightforward definitions for the
terms wersquoll be using to describe the ins and outs of text analytics solutions Feel free to check back in with these definitions as you read
Accuracy The combination of precision and
recall for a given tag or text analytics model
Emotion Measure of positivenegative
feelings strong and clear-cut enough to be
categorized as a specific emotion
Human Translation The method of having
each comment translated individually by
a human translator as it is submitted by
the customer
Intent Intent identifies what the customer is
trying to achieve based on their response
Keyword A word or term that occurs in
unstructured customer feedback data
Machine Translation Translation done by a
machine that has been trained by humans
Native Language Model A Text Analytics
model that is purposely built for a specific
spoken language
Natural Language Processing A field of
computer science amp artificial intelligence that
draws intelligence from unstructured data
Precision Correctness represents how
often a given concept is correctly captured
by a specific tag
Recall Coverage refers to how thoroughly the
topics or ideas within a given tag are captured
Sentiment The expressed feeling or atti-
tude behind a customerrsquos feedback cate-
gorized as positive negative or neutral
Sentiment Phrase Also referred to as a
Sentiment Bearing Phrase or SBP A phrase
or sentence identified with positive nega-
tive or neutral sentiment
Sentiment Score A measure for both the
polarity and intensity of the sentiment
within a given comment
Tag A label generated from text analytics
that groups together similar customer com-
ments around a specific concept or topic
Text Analytics The methods amp pro-
cesses used for obtaining insights from
unstructured data
Text Analytics Model A natural language
processing engine that uses tags to label
and organize unstructured data
Theme A dynamically extracted concept
from a collection of comments generated by
an unsupervised machine learning algorithm
Unstructured Data Qualitative data or
information that is not organized according
to an easily recognizable structure in-
cludes comments social data images or
audio recordings
04
05
05
LET rsquoS TALK ABOUT ACCURACY
What It Is and What It IsnrsquotNow that wersquove gone over the top terms itrsquos time to recognize a truism
vital to understanding text analytics as a whole
06
When we talk about analyzing unstructured data what wersquore really
asking computers to do is understand and analyze a very human
conceptmdashand then to do it in multiple languages
The truth is computers will never be able to understand 100 percent
of the nuances of human language (especially when you consider that
even humans often have trouble distinguishing and understanding
emotion sarcasm slang and all the other complexities inherent in
everyday speech) That is why it is so important to dispel common
misconceptions about accuracy Wersquove laid out a few fictions below
and replaced them with fact
The First Step to Unraveling the Mystery of Text Analytics Is to Understand That It Is Both an Art and a Science
07
DIFFERENTIATING ACCURACY FACT FROM FICTION
Fiction 1
ACCURACY HAS A SINGUL AR MEANING
The term ldquoaccuracyrdquo gets thrown around in text analytics
discussions a lot But what does it actually mean When
referring to the accuracy of the text analytics as a whole
accuracy is based on two factors recall and precision Often
CX vendors will tout 80-90 percent accuracy when what they
are really talking about is precision However what they do not
account for is recallmdashthe extent to which the model retrieves all
key terms or topics For example if the recall rate is only 60
you are missing 40 of actionable intelligence without even
realizing it
In addition when speaking to translation accuracy is often
confused with literalness or the degree of similarity between
exact words and grammar Accuracy is actually the degree of
similarity in meaning A comment does not need to be translated
literally to be accurate
False Negatives
TrueNegatives
Relevant Instances
Retr
ieve
d In
stan
ces
FalsePositives
TruePositives
PRECISION Fraction of retrieved instanced
that are relevant (how useful the results are)
RECALL Fraction of relevant instanced that
are retrieved (how complete the results are)
08
Fiction 3
ACCURACY IS THE GOAL
This may be the biggest misconception alive in text analytics
today Accuracy is not the goal of text analytics itrsquos a happy
byproduct The goal of text analytics in an experience program
is the ability to transform customer data into business
improvement Itrsquos important to never lose sight of that end goal
when designing your program and the text analytics solution that
helps to power it
Fiction 2
PERCENTAGES ARE A TRUE REPRESENTATION
OF ACCURACY
Yoursquove heard accuracy rates thrown around ldquoOur text analytics
are 85 accuraterdquo But have you ever thought about what that
actually means especially when yoursquore digging through your
data When you come across a mis-tagged or miscategorized
comment you might be tempted to think ldquoWell these text
analytics donrsquot workrdquo If you stop and think about it however if
the text analytics are 85 accurate that still means that 15
wonrsquot be And thatrsquos when percentages donrsquot always feel like
a true representation of accuracy Your brain might know that
15 of one thousand comments might mean 150 comments are
tagged incorrectly but when you see it in your data it can feel
like a lot more
08
09
09
TOP OBJECTIVES FOR A MODERN TEXT ANALY TICS SOLUTION
Scalability Quality Actionability and SpeedSo if 100 accuracy is neither achievable nor the end-all-be-all of text analytics solutions what
should experience programs strive for Here are four key areas to consider when measuring the
effectiveness of a text analytics solution
010
Scalability
Any and every growing business (especially those with
a global reach) needs a solution that supports all of the
countries and languages your customers work and buy
inmdashat an acceptable level of quality and at a price that
makes sense No one has unlimited resources so the
first criteria must be that the text analytics solution you
choose is scalable
Quality
The goal of text analytics is not to allow you to read
every single comment that your customers submit
For most businesses this is simply not feasible unless
you have a team of analysts dedicated to it full time
Your text analytics solution must be able to understand
and surface important trends and patterns based on
individual comments and the sentiments behind them
011
Actionability
Another key objective of text analytics is to be able to
take action on the customer issues that most affect your
business This means you need a layer of sophisticated
analytics that can add tags and themes on a granular level
uncover sentiment assign categories identify intent spot
legal issues and possible customer churn and more
Speed
Business today moves quickly so you need a text analytics
solution that can keep pacemdashwhich often means real-time
analysis and reporting This is specifically relevant when
considering translations for global companies The need
for speed means that brands donrsquot have time to wait for a
perfect translation of each and every comment because
perfect translations are only possible if a team manually
translates every word which takes a significant and
unrealistic amount of time amp money
Scalability
Any and every growing business (especially those with
a global reach) needs a solution that supports all of the
countries and languages your customers work and buy
inmdashat an acceptable level of quality and at a price that
makes sense No one has unlimited resources so the
first criteria must be that the text analytics solution you
choose is scalable
012
Tiffany amp Co Leverages Global Text Analytics for Market-Specific and Company-Wide Understanding of Experiences
For a global company the ability to collect and analyze customer feedback across multiple regions and languagesmdash
within a single platformmdashis essential Tiffany amp Co one of the worldrsquos most recognizable and respected luxury
brands leverages InMoment in 18 languages across 300 company-owned stores multiple authorized retailers
its e-commerce platform and in its customer care center for a real-time understanding of market-specific and
company-wide feedback trends
After translation to English comments are analyzed using approximately 150 custom tags and customer sentiment
is displayed in a multitude of reports and dashboards in both English and the original comment language This
way both local and corporate program owners can view results in the language that is most comfortable for them
Further InMomentrsquos highly accurate translation means Tiffany amp Co may use any and every platform featuremdashon
feedback collected in all languagesmdashto sift through comments and research topics most important to the brand
With more than 15 million pieces of customer feedback collected each year comments are used to recognize
exceptional locations and employees identify areas of opportunity and understand the specific drivers of customer
satisfaction The ability to analyze feedback across 18 different languages helps Tiffany amp Co understand nuances
in the experiences and expectations of its customers across each region market and location
013
13
GET TING GLOBAL WITH TRANSL ATION
Human Machine and MoreAs we mentioned in the last section the translation method you choose
can affect the goals you set for the scalability quality actionability and
speed of your text analytics solution
014
Human TranslationHuman translation is the method of having each comment translated
individually as it is submitted by the customer This type of manual
translation requires a long expensive process
Instead of real-time results there is usually a wait of 24 to 48 hours (or
more) to get the translated results And because human translation is
limited to languages that a specific translator can work with outsourcing
more complex languages to other translators will also add additional
costsmdashand more time
LETrsquoS GO OVER THE DIFFERENT OPTIONS
AND THEIR PROS AND CONS
015
Native Language LibrariesA native language library is a text analytics model created
entirely in a language of choice so the system analyzes and
tags comments in that native language instead of translating
them to English tagging them and exporting them back to the
original language Creating a native language library involves a
native speaker building out the initial text analytics model and
then constantly refining it to ensure that it remains up to date
Creating language libraries is a slow and expensive process
usually costing hundreds of thousands of dollars and taking
months or years to build Large amounts of data must be
collected collated and categorized in the language of choice
before the libraries can be built In addition each of these
libraries must be built and tuned to individual industries to
ensure their accuracy The main problem with these native
language libraries therefore is the lack of scalability
016
Machine Translated Human LedMachine translation is exactly what it sounds like It uses
machinesmdashtrained and monitored by humansmdashto translate
text When translating machine translation services look for
patterns and trends in millions of documents to provide the
best translation
Additionally many machine translation services are now using
a neural machine translation engine This means that rather
than translating individual words it uses a neural network
capable of deep learning to look at whole sentences and
understand the greater context These translation services are
highly accurate actionable cost efficient and scalable which
makes them the best option for an international VoC program
017
17
A ldquoMODELrdquo TEXT ANALY TICS SOLUTION
Tailoring Intelligence for Your BusinessThe next element of a text analytics solution to explore is the models you can employ to label
and organize your unstructured data These models have truly come a long way when it comes to their ability to deliver intelligence that is specifically tailored to your business so you can identify
areas that require action and deliver direct value to your bottom line
Traditionally brands were faced with the choice between two specific types of text analytics models industry models and custom models The definition of these two historical approaches
is self explanatory but letrsquos take a deeper dive
018
Industry ModelsAn industry model is developed by experts to reflect the
characteristics of a specific industry These models are
often designed by a vendor to be employed for every client
that fits that particular industry and therefore it often
misses brand-specific insights Additionally because the
same model is used for multiple brands the model canrsquot
be adapted for one brand without being adapted for all
However these models are a much more time-and-cost
effective solution when compared to custom models
Custom ModelsA custom model is built specifically for a certain company
gaining individualized context on the moments that set
a brand apart from its competitors While they are highly
accurate these models are more expensive to create and
maintain so they arenrsquot updated regularly Additionally they
require a high number of comments to test
019
Custom Layered ModelsThere are pros and cons to both of these historical
approaches but there is a new more effective
approach on the scene that gives experience
programs the best of both worlds custom layered
models Custom layered models take the accuracy of
custom models and the efficiency of industry models
to provide tailored accurate and effective results at a
fraction of the cost
How InMomentrsquos Custom Layered Models leverage
over 50 industry models to craft a personalized
tailored solution for each brand And when your text
analytics are built to suit your specific needs the
intelligence you gain from them is multiplied Take a
look at the difference
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retailecommerce
pharmacycustom tags
3xMore Data Points
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retail
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
SINGLE INDUSTRY MODEL CUSTOM L AYERED MODEL
019
020
Comparing and Contrasting ModelsWant a straightforward look at how historical industry and custom models compare to custom layed models
Wersquove laid out the features and benefits of each in this chart
INDUSTRY MODELS CUSTOM MODELS CUSTOM LAYERED MODELS
Covers All Industries
Reflects Unique Business
Delivers Insights for a Combination of Offerings Channels amp Services
Can Be Adapted for an Individual Business without Changing the Model for All Clients Using That Model
Highly Accurate
Cost Effective
Time Efficient
Updated Regularly
Can Be Tested with Limited Data
020
021
21
FACTORING IN SENTIMENT
Getting to the WhyWhen humans look to tell effective storiesmdashwhether in film or in conversationmdashsentiment is
absolutely essential Itrsquos no different when your customer or employee is trying to tell you a
story about their experience via unstructured feedback Itrsquos your text analytics solutionrsquos job to
detect the emotion in their story identify it and help you understand what is causing that specific
sentiment so you can improve experiences But the specific method you use to help your solution
detect sentiment can affect the results you get and therefore your ability to act and transform
022
Rules-Based SentimentTraditionally the way text analytics identified senti-
ment was through a rules-based approach As the name
suggests this method applied rigid rules to understand
unstructured feedback Additionally changing these
rules or updating them required manual intervention
And even with those manual updates accuracy never
really improves
Adaptive Sentiment EngineThe idea that inspired the adaptive sentiment engine
approach is that rigid rules do not lend themselves to best
understand human emotions Instead adaptive sentiment
is AI driven meaning that it learns and improves over time
giving you more accurate results This engine continually
recognizes new terms and phrases as well meaning that
it grows and adapts in lockstep with your brand Check out
the difference below
023
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive Positive Negative NegativeSentiment Score 92
Confidence 98Sentiment Score 52
Confidence 97Sentiment Score -49
Confidence 41Sentiment Score -79
Confidence 94
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
RULES -BASED MACHINE LEARNING
NegativeNeutral
024
Now that you know what it takes to make your text analytics
solution truly world-class what next Well itrsquos time to make
sure that your greater experience program is set up to actually
drive improvement for your organization
Text analytics are just one piece of the puzzle when it comes
to using your VoC VoE or other experience initiatives to
power business growth When you look at the bigger picture
it becomes clear that you need a solid strategy for success
At InMoment we leverage the Continuous Improvement
Framework a method that helps us guide our clients through
designing listening understanding transforming and realizing
their goals for their experience initiative
If you want to learn more about this framework
and how text analytics fit into it you can read
more here
Like what yoursquove seen so far
Here are a few other pieces you may like
How You Listen Matters Modernizing Your Methods amp Approach to Customer Feedback
How to Prove the Business Value of Your CX Program
Building a Business Case for Your CFO (or Anyone Else)
Measuring Doesnrsquot Make You Taller
Therersquos more where that came from on
wwwinmomentcomresources
DRIVING IMPROVEMENT
024
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C
02
IntroductionText analytics are vital to your brandrsquos ability to understand your
customer and employee experiences You can have listening posts
across every channel and at every point in the customer journey
but if you donrsquot have the best possible text analytics solution in
place your ability to derive actionable intelligence from that data
is essentially moot And your ability to create transformational
change across the organization and drive business growth Thatrsquod
be a non-starter without effective text analytics Without them
all you have is a score not any context or information on what
actually went well or needs improvement
Itrsquos obvious that text analytics are vital but in an industry full of
jargon claims about accuracy and a huge amount of conflicting
data how can you tell what solution attributes will be the best for
your company
Well in this eBook wersquore cutting through the fancy lingo and
salesy claims to get to the heart of what makes a text analytics
solution truly successful Keep reading to learn more about top
terms accuracy solution necessities and more
Without text analytics all you have is a score not any context or information on what actually went well or needs improvement
03
TABL
E O
F CO
NTE
NTS Pg 04 Pg 05 Pg 09
Pg 21Pg 17Pg 13
Pg 24
TEXT ANALYTICS 101
A QUICK GLOSSARY
A ldquoMODELrdquo TEXT
ANALYTICS SOLUTIONFACTORING IN SENTIMENTGETTING GLOBAL
WITH TRANSLATION
DRIVING IMPROVEMENT
LETrsquoS TALK ABOUT
ACCURACY
TOP OBJECTIVES FOR A
MODERN TA SOLUTION
04TEXT ANALY TICS 101A Quick Glossary
Therersquos no way around it conversations around text analytics are chock-full of jargon Below yoursquoll find straightforward definitions for the
terms wersquoll be using to describe the ins and outs of text analytics solutions Feel free to check back in with these definitions as you read
Accuracy The combination of precision and
recall for a given tag or text analytics model
Emotion Measure of positivenegative
feelings strong and clear-cut enough to be
categorized as a specific emotion
Human Translation The method of having
each comment translated individually by
a human translator as it is submitted by
the customer
Intent Intent identifies what the customer is
trying to achieve based on their response
Keyword A word or term that occurs in
unstructured customer feedback data
Machine Translation Translation done by a
machine that has been trained by humans
Native Language Model A Text Analytics
model that is purposely built for a specific
spoken language
Natural Language Processing A field of
computer science amp artificial intelligence that
draws intelligence from unstructured data
Precision Correctness represents how
often a given concept is correctly captured
by a specific tag
Recall Coverage refers to how thoroughly the
topics or ideas within a given tag are captured
Sentiment The expressed feeling or atti-
tude behind a customerrsquos feedback cate-
gorized as positive negative or neutral
Sentiment Phrase Also referred to as a
Sentiment Bearing Phrase or SBP A phrase
or sentence identified with positive nega-
tive or neutral sentiment
Sentiment Score A measure for both the
polarity and intensity of the sentiment
within a given comment
Tag A label generated from text analytics
that groups together similar customer com-
ments around a specific concept or topic
Text Analytics The methods amp pro-
cesses used for obtaining insights from
unstructured data
Text Analytics Model A natural language
processing engine that uses tags to label
and organize unstructured data
Theme A dynamically extracted concept
from a collection of comments generated by
an unsupervised machine learning algorithm
Unstructured Data Qualitative data or
information that is not organized according
to an easily recognizable structure in-
cludes comments social data images or
audio recordings
04
05
05
LET rsquoS TALK ABOUT ACCURACY
What It Is and What It IsnrsquotNow that wersquove gone over the top terms itrsquos time to recognize a truism
vital to understanding text analytics as a whole
06
When we talk about analyzing unstructured data what wersquore really
asking computers to do is understand and analyze a very human
conceptmdashand then to do it in multiple languages
The truth is computers will never be able to understand 100 percent
of the nuances of human language (especially when you consider that
even humans often have trouble distinguishing and understanding
emotion sarcasm slang and all the other complexities inherent in
everyday speech) That is why it is so important to dispel common
misconceptions about accuracy Wersquove laid out a few fictions below
and replaced them with fact
The First Step to Unraveling the Mystery of Text Analytics Is to Understand That It Is Both an Art and a Science
07
DIFFERENTIATING ACCURACY FACT FROM FICTION
Fiction 1
ACCURACY HAS A SINGUL AR MEANING
The term ldquoaccuracyrdquo gets thrown around in text analytics
discussions a lot But what does it actually mean When
referring to the accuracy of the text analytics as a whole
accuracy is based on two factors recall and precision Often
CX vendors will tout 80-90 percent accuracy when what they
are really talking about is precision However what they do not
account for is recallmdashthe extent to which the model retrieves all
key terms or topics For example if the recall rate is only 60
you are missing 40 of actionable intelligence without even
realizing it
In addition when speaking to translation accuracy is often
confused with literalness or the degree of similarity between
exact words and grammar Accuracy is actually the degree of
similarity in meaning A comment does not need to be translated
literally to be accurate
False Negatives
TrueNegatives
Relevant Instances
Retr
ieve
d In
stan
ces
FalsePositives
TruePositives
PRECISION Fraction of retrieved instanced
that are relevant (how useful the results are)
RECALL Fraction of relevant instanced that
are retrieved (how complete the results are)
08
Fiction 3
ACCURACY IS THE GOAL
This may be the biggest misconception alive in text analytics
today Accuracy is not the goal of text analytics itrsquos a happy
byproduct The goal of text analytics in an experience program
is the ability to transform customer data into business
improvement Itrsquos important to never lose sight of that end goal
when designing your program and the text analytics solution that
helps to power it
Fiction 2
PERCENTAGES ARE A TRUE REPRESENTATION
OF ACCURACY
Yoursquove heard accuracy rates thrown around ldquoOur text analytics
are 85 accuraterdquo But have you ever thought about what that
actually means especially when yoursquore digging through your
data When you come across a mis-tagged or miscategorized
comment you might be tempted to think ldquoWell these text
analytics donrsquot workrdquo If you stop and think about it however if
the text analytics are 85 accurate that still means that 15
wonrsquot be And thatrsquos when percentages donrsquot always feel like
a true representation of accuracy Your brain might know that
15 of one thousand comments might mean 150 comments are
tagged incorrectly but when you see it in your data it can feel
like a lot more
08
09
09
TOP OBJECTIVES FOR A MODERN TEXT ANALY TICS SOLUTION
Scalability Quality Actionability and SpeedSo if 100 accuracy is neither achievable nor the end-all-be-all of text analytics solutions what
should experience programs strive for Here are four key areas to consider when measuring the
effectiveness of a text analytics solution
010
Scalability
Any and every growing business (especially those with
a global reach) needs a solution that supports all of the
countries and languages your customers work and buy
inmdashat an acceptable level of quality and at a price that
makes sense No one has unlimited resources so the
first criteria must be that the text analytics solution you
choose is scalable
Quality
The goal of text analytics is not to allow you to read
every single comment that your customers submit
For most businesses this is simply not feasible unless
you have a team of analysts dedicated to it full time
Your text analytics solution must be able to understand
and surface important trends and patterns based on
individual comments and the sentiments behind them
011
Actionability
Another key objective of text analytics is to be able to
take action on the customer issues that most affect your
business This means you need a layer of sophisticated
analytics that can add tags and themes on a granular level
uncover sentiment assign categories identify intent spot
legal issues and possible customer churn and more
Speed
Business today moves quickly so you need a text analytics
solution that can keep pacemdashwhich often means real-time
analysis and reporting This is specifically relevant when
considering translations for global companies The need
for speed means that brands donrsquot have time to wait for a
perfect translation of each and every comment because
perfect translations are only possible if a team manually
translates every word which takes a significant and
unrealistic amount of time amp money
Scalability
Any and every growing business (especially those with
a global reach) needs a solution that supports all of the
countries and languages your customers work and buy
inmdashat an acceptable level of quality and at a price that
makes sense No one has unlimited resources so the
first criteria must be that the text analytics solution you
choose is scalable
012
Tiffany amp Co Leverages Global Text Analytics for Market-Specific and Company-Wide Understanding of Experiences
For a global company the ability to collect and analyze customer feedback across multiple regions and languagesmdash
within a single platformmdashis essential Tiffany amp Co one of the worldrsquos most recognizable and respected luxury
brands leverages InMoment in 18 languages across 300 company-owned stores multiple authorized retailers
its e-commerce platform and in its customer care center for a real-time understanding of market-specific and
company-wide feedback trends
After translation to English comments are analyzed using approximately 150 custom tags and customer sentiment
is displayed in a multitude of reports and dashboards in both English and the original comment language This
way both local and corporate program owners can view results in the language that is most comfortable for them
Further InMomentrsquos highly accurate translation means Tiffany amp Co may use any and every platform featuremdashon
feedback collected in all languagesmdashto sift through comments and research topics most important to the brand
With more than 15 million pieces of customer feedback collected each year comments are used to recognize
exceptional locations and employees identify areas of opportunity and understand the specific drivers of customer
satisfaction The ability to analyze feedback across 18 different languages helps Tiffany amp Co understand nuances
in the experiences and expectations of its customers across each region market and location
013
13
GET TING GLOBAL WITH TRANSL ATION
Human Machine and MoreAs we mentioned in the last section the translation method you choose
can affect the goals you set for the scalability quality actionability and
speed of your text analytics solution
014
Human TranslationHuman translation is the method of having each comment translated
individually as it is submitted by the customer This type of manual
translation requires a long expensive process
Instead of real-time results there is usually a wait of 24 to 48 hours (or
more) to get the translated results And because human translation is
limited to languages that a specific translator can work with outsourcing
more complex languages to other translators will also add additional
costsmdashand more time
LETrsquoS GO OVER THE DIFFERENT OPTIONS
AND THEIR PROS AND CONS
015
Native Language LibrariesA native language library is a text analytics model created
entirely in a language of choice so the system analyzes and
tags comments in that native language instead of translating
them to English tagging them and exporting them back to the
original language Creating a native language library involves a
native speaker building out the initial text analytics model and
then constantly refining it to ensure that it remains up to date
Creating language libraries is a slow and expensive process
usually costing hundreds of thousands of dollars and taking
months or years to build Large amounts of data must be
collected collated and categorized in the language of choice
before the libraries can be built In addition each of these
libraries must be built and tuned to individual industries to
ensure their accuracy The main problem with these native
language libraries therefore is the lack of scalability
016
Machine Translated Human LedMachine translation is exactly what it sounds like It uses
machinesmdashtrained and monitored by humansmdashto translate
text When translating machine translation services look for
patterns and trends in millions of documents to provide the
best translation
Additionally many machine translation services are now using
a neural machine translation engine This means that rather
than translating individual words it uses a neural network
capable of deep learning to look at whole sentences and
understand the greater context These translation services are
highly accurate actionable cost efficient and scalable which
makes them the best option for an international VoC program
017
17
A ldquoMODELrdquo TEXT ANALY TICS SOLUTION
Tailoring Intelligence for Your BusinessThe next element of a text analytics solution to explore is the models you can employ to label
and organize your unstructured data These models have truly come a long way when it comes to their ability to deliver intelligence that is specifically tailored to your business so you can identify
areas that require action and deliver direct value to your bottom line
Traditionally brands were faced with the choice between two specific types of text analytics models industry models and custom models The definition of these two historical approaches
is self explanatory but letrsquos take a deeper dive
018
Industry ModelsAn industry model is developed by experts to reflect the
characteristics of a specific industry These models are
often designed by a vendor to be employed for every client
that fits that particular industry and therefore it often
misses brand-specific insights Additionally because the
same model is used for multiple brands the model canrsquot
be adapted for one brand without being adapted for all
However these models are a much more time-and-cost
effective solution when compared to custom models
Custom ModelsA custom model is built specifically for a certain company
gaining individualized context on the moments that set
a brand apart from its competitors While they are highly
accurate these models are more expensive to create and
maintain so they arenrsquot updated regularly Additionally they
require a high number of comments to test
019
Custom Layered ModelsThere are pros and cons to both of these historical
approaches but there is a new more effective
approach on the scene that gives experience
programs the best of both worlds custom layered
models Custom layered models take the accuracy of
custom models and the efficiency of industry models
to provide tailored accurate and effective results at a
fraction of the cost
How InMomentrsquos Custom Layered Models leverage
over 50 industry models to craft a personalized
tailored solution for each brand And when your text
analytics are built to suit your specific needs the
intelligence you gain from them is multiplied Take a
look at the difference
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retailecommerce
pharmacycustom tags
3xMore Data Points
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retail
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
SINGLE INDUSTRY MODEL CUSTOM L AYERED MODEL
019
020
Comparing and Contrasting ModelsWant a straightforward look at how historical industry and custom models compare to custom layed models
Wersquove laid out the features and benefits of each in this chart
INDUSTRY MODELS CUSTOM MODELS CUSTOM LAYERED MODELS
Covers All Industries
Reflects Unique Business
Delivers Insights for a Combination of Offerings Channels amp Services
Can Be Adapted for an Individual Business without Changing the Model for All Clients Using That Model
Highly Accurate
Cost Effective
Time Efficient
Updated Regularly
Can Be Tested with Limited Data
020
021
21
FACTORING IN SENTIMENT
Getting to the WhyWhen humans look to tell effective storiesmdashwhether in film or in conversationmdashsentiment is
absolutely essential Itrsquos no different when your customer or employee is trying to tell you a
story about their experience via unstructured feedback Itrsquos your text analytics solutionrsquos job to
detect the emotion in their story identify it and help you understand what is causing that specific
sentiment so you can improve experiences But the specific method you use to help your solution
detect sentiment can affect the results you get and therefore your ability to act and transform
022
Rules-Based SentimentTraditionally the way text analytics identified senti-
ment was through a rules-based approach As the name
suggests this method applied rigid rules to understand
unstructured feedback Additionally changing these
rules or updating them required manual intervention
And even with those manual updates accuracy never
really improves
Adaptive Sentiment EngineThe idea that inspired the adaptive sentiment engine
approach is that rigid rules do not lend themselves to best
understand human emotions Instead adaptive sentiment
is AI driven meaning that it learns and improves over time
giving you more accurate results This engine continually
recognizes new terms and phrases as well meaning that
it grows and adapts in lockstep with your brand Check out
the difference below
023
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive Positive Negative NegativeSentiment Score 92
Confidence 98Sentiment Score 52
Confidence 97Sentiment Score -49
Confidence 41Sentiment Score -79
Confidence 94
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
RULES -BASED MACHINE LEARNING
NegativeNeutral
024
Now that you know what it takes to make your text analytics
solution truly world-class what next Well itrsquos time to make
sure that your greater experience program is set up to actually
drive improvement for your organization
Text analytics are just one piece of the puzzle when it comes
to using your VoC VoE or other experience initiatives to
power business growth When you look at the bigger picture
it becomes clear that you need a solid strategy for success
At InMoment we leverage the Continuous Improvement
Framework a method that helps us guide our clients through
designing listening understanding transforming and realizing
their goals for their experience initiative
If you want to learn more about this framework
and how text analytics fit into it you can read
more here
Like what yoursquove seen so far
Here are a few other pieces you may like
How You Listen Matters Modernizing Your Methods amp Approach to Customer Feedback
How to Prove the Business Value of Your CX Program
Building a Business Case for Your CFO (or Anyone Else)
Measuring Doesnrsquot Make You Taller
Therersquos more where that came from on
wwwinmomentcomresources
DRIVING IMPROVEMENT
024
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C
03
TABL
E O
F CO
NTE
NTS Pg 04 Pg 05 Pg 09
Pg 21Pg 17Pg 13
Pg 24
TEXT ANALYTICS 101
A QUICK GLOSSARY
A ldquoMODELrdquo TEXT
ANALYTICS SOLUTIONFACTORING IN SENTIMENTGETTING GLOBAL
WITH TRANSLATION
DRIVING IMPROVEMENT
LETrsquoS TALK ABOUT
ACCURACY
TOP OBJECTIVES FOR A
MODERN TA SOLUTION
04TEXT ANALY TICS 101A Quick Glossary
Therersquos no way around it conversations around text analytics are chock-full of jargon Below yoursquoll find straightforward definitions for the
terms wersquoll be using to describe the ins and outs of text analytics solutions Feel free to check back in with these definitions as you read
Accuracy The combination of precision and
recall for a given tag or text analytics model
Emotion Measure of positivenegative
feelings strong and clear-cut enough to be
categorized as a specific emotion
Human Translation The method of having
each comment translated individually by
a human translator as it is submitted by
the customer
Intent Intent identifies what the customer is
trying to achieve based on their response
Keyword A word or term that occurs in
unstructured customer feedback data
Machine Translation Translation done by a
machine that has been trained by humans
Native Language Model A Text Analytics
model that is purposely built for a specific
spoken language
Natural Language Processing A field of
computer science amp artificial intelligence that
draws intelligence from unstructured data
Precision Correctness represents how
often a given concept is correctly captured
by a specific tag
Recall Coverage refers to how thoroughly the
topics or ideas within a given tag are captured
Sentiment The expressed feeling or atti-
tude behind a customerrsquos feedback cate-
gorized as positive negative or neutral
Sentiment Phrase Also referred to as a
Sentiment Bearing Phrase or SBP A phrase
or sentence identified with positive nega-
tive or neutral sentiment
Sentiment Score A measure for both the
polarity and intensity of the sentiment
within a given comment
Tag A label generated from text analytics
that groups together similar customer com-
ments around a specific concept or topic
Text Analytics The methods amp pro-
cesses used for obtaining insights from
unstructured data
Text Analytics Model A natural language
processing engine that uses tags to label
and organize unstructured data
Theme A dynamically extracted concept
from a collection of comments generated by
an unsupervised machine learning algorithm
Unstructured Data Qualitative data or
information that is not organized according
to an easily recognizable structure in-
cludes comments social data images or
audio recordings
04
05
05
LET rsquoS TALK ABOUT ACCURACY
What It Is and What It IsnrsquotNow that wersquove gone over the top terms itrsquos time to recognize a truism
vital to understanding text analytics as a whole
06
When we talk about analyzing unstructured data what wersquore really
asking computers to do is understand and analyze a very human
conceptmdashand then to do it in multiple languages
The truth is computers will never be able to understand 100 percent
of the nuances of human language (especially when you consider that
even humans often have trouble distinguishing and understanding
emotion sarcasm slang and all the other complexities inherent in
everyday speech) That is why it is so important to dispel common
misconceptions about accuracy Wersquove laid out a few fictions below
and replaced them with fact
The First Step to Unraveling the Mystery of Text Analytics Is to Understand That It Is Both an Art and a Science
07
DIFFERENTIATING ACCURACY FACT FROM FICTION
Fiction 1
ACCURACY HAS A SINGUL AR MEANING
The term ldquoaccuracyrdquo gets thrown around in text analytics
discussions a lot But what does it actually mean When
referring to the accuracy of the text analytics as a whole
accuracy is based on two factors recall and precision Often
CX vendors will tout 80-90 percent accuracy when what they
are really talking about is precision However what they do not
account for is recallmdashthe extent to which the model retrieves all
key terms or topics For example if the recall rate is only 60
you are missing 40 of actionable intelligence without even
realizing it
In addition when speaking to translation accuracy is often
confused with literalness or the degree of similarity between
exact words and grammar Accuracy is actually the degree of
similarity in meaning A comment does not need to be translated
literally to be accurate
False Negatives
TrueNegatives
Relevant Instances
Retr
ieve
d In
stan
ces
FalsePositives
TruePositives
PRECISION Fraction of retrieved instanced
that are relevant (how useful the results are)
RECALL Fraction of relevant instanced that
are retrieved (how complete the results are)
08
Fiction 3
ACCURACY IS THE GOAL
This may be the biggest misconception alive in text analytics
today Accuracy is not the goal of text analytics itrsquos a happy
byproduct The goal of text analytics in an experience program
is the ability to transform customer data into business
improvement Itrsquos important to never lose sight of that end goal
when designing your program and the text analytics solution that
helps to power it
Fiction 2
PERCENTAGES ARE A TRUE REPRESENTATION
OF ACCURACY
Yoursquove heard accuracy rates thrown around ldquoOur text analytics
are 85 accuraterdquo But have you ever thought about what that
actually means especially when yoursquore digging through your
data When you come across a mis-tagged or miscategorized
comment you might be tempted to think ldquoWell these text
analytics donrsquot workrdquo If you stop and think about it however if
the text analytics are 85 accurate that still means that 15
wonrsquot be And thatrsquos when percentages donrsquot always feel like
a true representation of accuracy Your brain might know that
15 of one thousand comments might mean 150 comments are
tagged incorrectly but when you see it in your data it can feel
like a lot more
08
09
09
TOP OBJECTIVES FOR A MODERN TEXT ANALY TICS SOLUTION
Scalability Quality Actionability and SpeedSo if 100 accuracy is neither achievable nor the end-all-be-all of text analytics solutions what
should experience programs strive for Here are four key areas to consider when measuring the
effectiveness of a text analytics solution
010
Scalability
Any and every growing business (especially those with
a global reach) needs a solution that supports all of the
countries and languages your customers work and buy
inmdashat an acceptable level of quality and at a price that
makes sense No one has unlimited resources so the
first criteria must be that the text analytics solution you
choose is scalable
Quality
The goal of text analytics is not to allow you to read
every single comment that your customers submit
For most businesses this is simply not feasible unless
you have a team of analysts dedicated to it full time
Your text analytics solution must be able to understand
and surface important trends and patterns based on
individual comments and the sentiments behind them
011
Actionability
Another key objective of text analytics is to be able to
take action on the customer issues that most affect your
business This means you need a layer of sophisticated
analytics that can add tags and themes on a granular level
uncover sentiment assign categories identify intent spot
legal issues and possible customer churn and more
Speed
Business today moves quickly so you need a text analytics
solution that can keep pacemdashwhich often means real-time
analysis and reporting This is specifically relevant when
considering translations for global companies The need
for speed means that brands donrsquot have time to wait for a
perfect translation of each and every comment because
perfect translations are only possible if a team manually
translates every word which takes a significant and
unrealistic amount of time amp money
Scalability
Any and every growing business (especially those with
a global reach) needs a solution that supports all of the
countries and languages your customers work and buy
inmdashat an acceptable level of quality and at a price that
makes sense No one has unlimited resources so the
first criteria must be that the text analytics solution you
choose is scalable
012
Tiffany amp Co Leverages Global Text Analytics for Market-Specific and Company-Wide Understanding of Experiences
For a global company the ability to collect and analyze customer feedback across multiple regions and languagesmdash
within a single platformmdashis essential Tiffany amp Co one of the worldrsquos most recognizable and respected luxury
brands leverages InMoment in 18 languages across 300 company-owned stores multiple authorized retailers
its e-commerce platform and in its customer care center for a real-time understanding of market-specific and
company-wide feedback trends
After translation to English comments are analyzed using approximately 150 custom tags and customer sentiment
is displayed in a multitude of reports and dashboards in both English and the original comment language This
way both local and corporate program owners can view results in the language that is most comfortable for them
Further InMomentrsquos highly accurate translation means Tiffany amp Co may use any and every platform featuremdashon
feedback collected in all languagesmdashto sift through comments and research topics most important to the brand
With more than 15 million pieces of customer feedback collected each year comments are used to recognize
exceptional locations and employees identify areas of opportunity and understand the specific drivers of customer
satisfaction The ability to analyze feedback across 18 different languages helps Tiffany amp Co understand nuances
in the experiences and expectations of its customers across each region market and location
013
13
GET TING GLOBAL WITH TRANSL ATION
Human Machine and MoreAs we mentioned in the last section the translation method you choose
can affect the goals you set for the scalability quality actionability and
speed of your text analytics solution
014
Human TranslationHuman translation is the method of having each comment translated
individually as it is submitted by the customer This type of manual
translation requires a long expensive process
Instead of real-time results there is usually a wait of 24 to 48 hours (or
more) to get the translated results And because human translation is
limited to languages that a specific translator can work with outsourcing
more complex languages to other translators will also add additional
costsmdashand more time
LETrsquoS GO OVER THE DIFFERENT OPTIONS
AND THEIR PROS AND CONS
015
Native Language LibrariesA native language library is a text analytics model created
entirely in a language of choice so the system analyzes and
tags comments in that native language instead of translating
them to English tagging them and exporting them back to the
original language Creating a native language library involves a
native speaker building out the initial text analytics model and
then constantly refining it to ensure that it remains up to date
Creating language libraries is a slow and expensive process
usually costing hundreds of thousands of dollars and taking
months or years to build Large amounts of data must be
collected collated and categorized in the language of choice
before the libraries can be built In addition each of these
libraries must be built and tuned to individual industries to
ensure their accuracy The main problem with these native
language libraries therefore is the lack of scalability
016
Machine Translated Human LedMachine translation is exactly what it sounds like It uses
machinesmdashtrained and monitored by humansmdashto translate
text When translating machine translation services look for
patterns and trends in millions of documents to provide the
best translation
Additionally many machine translation services are now using
a neural machine translation engine This means that rather
than translating individual words it uses a neural network
capable of deep learning to look at whole sentences and
understand the greater context These translation services are
highly accurate actionable cost efficient and scalable which
makes them the best option for an international VoC program
017
17
A ldquoMODELrdquo TEXT ANALY TICS SOLUTION
Tailoring Intelligence for Your BusinessThe next element of a text analytics solution to explore is the models you can employ to label
and organize your unstructured data These models have truly come a long way when it comes to their ability to deliver intelligence that is specifically tailored to your business so you can identify
areas that require action and deliver direct value to your bottom line
Traditionally brands were faced with the choice between two specific types of text analytics models industry models and custom models The definition of these two historical approaches
is self explanatory but letrsquos take a deeper dive
018
Industry ModelsAn industry model is developed by experts to reflect the
characteristics of a specific industry These models are
often designed by a vendor to be employed for every client
that fits that particular industry and therefore it often
misses brand-specific insights Additionally because the
same model is used for multiple brands the model canrsquot
be adapted for one brand without being adapted for all
However these models are a much more time-and-cost
effective solution when compared to custom models
Custom ModelsA custom model is built specifically for a certain company
gaining individualized context on the moments that set
a brand apart from its competitors While they are highly
accurate these models are more expensive to create and
maintain so they arenrsquot updated regularly Additionally they
require a high number of comments to test
019
Custom Layered ModelsThere are pros and cons to both of these historical
approaches but there is a new more effective
approach on the scene that gives experience
programs the best of both worlds custom layered
models Custom layered models take the accuracy of
custom models and the efficiency of industry models
to provide tailored accurate and effective results at a
fraction of the cost
How InMomentrsquos Custom Layered Models leverage
over 50 industry models to craft a personalized
tailored solution for each brand And when your text
analytics are built to suit your specific needs the
intelligence you gain from them is multiplied Take a
look at the difference
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retailecommerce
pharmacycustom tags
3xMore Data Points
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retail
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
SINGLE INDUSTRY MODEL CUSTOM L AYERED MODEL
019
020
Comparing and Contrasting ModelsWant a straightforward look at how historical industry and custom models compare to custom layed models
Wersquove laid out the features and benefits of each in this chart
INDUSTRY MODELS CUSTOM MODELS CUSTOM LAYERED MODELS
Covers All Industries
Reflects Unique Business
Delivers Insights for a Combination of Offerings Channels amp Services
Can Be Adapted for an Individual Business without Changing the Model for All Clients Using That Model
Highly Accurate
Cost Effective
Time Efficient
Updated Regularly
Can Be Tested with Limited Data
020
021
21
FACTORING IN SENTIMENT
Getting to the WhyWhen humans look to tell effective storiesmdashwhether in film or in conversationmdashsentiment is
absolutely essential Itrsquos no different when your customer or employee is trying to tell you a
story about their experience via unstructured feedback Itrsquos your text analytics solutionrsquos job to
detect the emotion in their story identify it and help you understand what is causing that specific
sentiment so you can improve experiences But the specific method you use to help your solution
detect sentiment can affect the results you get and therefore your ability to act and transform
022
Rules-Based SentimentTraditionally the way text analytics identified senti-
ment was through a rules-based approach As the name
suggests this method applied rigid rules to understand
unstructured feedback Additionally changing these
rules or updating them required manual intervention
And even with those manual updates accuracy never
really improves
Adaptive Sentiment EngineThe idea that inspired the adaptive sentiment engine
approach is that rigid rules do not lend themselves to best
understand human emotions Instead adaptive sentiment
is AI driven meaning that it learns and improves over time
giving you more accurate results This engine continually
recognizes new terms and phrases as well meaning that
it grows and adapts in lockstep with your brand Check out
the difference below
023
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive Positive Negative NegativeSentiment Score 92
Confidence 98Sentiment Score 52
Confidence 97Sentiment Score -49
Confidence 41Sentiment Score -79
Confidence 94
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
RULES -BASED MACHINE LEARNING
NegativeNeutral
024
Now that you know what it takes to make your text analytics
solution truly world-class what next Well itrsquos time to make
sure that your greater experience program is set up to actually
drive improvement for your organization
Text analytics are just one piece of the puzzle when it comes
to using your VoC VoE or other experience initiatives to
power business growth When you look at the bigger picture
it becomes clear that you need a solid strategy for success
At InMoment we leverage the Continuous Improvement
Framework a method that helps us guide our clients through
designing listening understanding transforming and realizing
their goals for their experience initiative
If you want to learn more about this framework
and how text analytics fit into it you can read
more here
Like what yoursquove seen so far
Here are a few other pieces you may like
How You Listen Matters Modernizing Your Methods amp Approach to Customer Feedback
How to Prove the Business Value of Your CX Program
Building a Business Case for Your CFO (or Anyone Else)
Measuring Doesnrsquot Make You Taller
Therersquos more where that came from on
wwwinmomentcomresources
DRIVING IMPROVEMENT
024
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C
04TEXT ANALY TICS 101A Quick Glossary
Therersquos no way around it conversations around text analytics are chock-full of jargon Below yoursquoll find straightforward definitions for the
terms wersquoll be using to describe the ins and outs of text analytics solutions Feel free to check back in with these definitions as you read
Accuracy The combination of precision and
recall for a given tag or text analytics model
Emotion Measure of positivenegative
feelings strong and clear-cut enough to be
categorized as a specific emotion
Human Translation The method of having
each comment translated individually by
a human translator as it is submitted by
the customer
Intent Intent identifies what the customer is
trying to achieve based on their response
Keyword A word or term that occurs in
unstructured customer feedback data
Machine Translation Translation done by a
machine that has been trained by humans
Native Language Model A Text Analytics
model that is purposely built for a specific
spoken language
Natural Language Processing A field of
computer science amp artificial intelligence that
draws intelligence from unstructured data
Precision Correctness represents how
often a given concept is correctly captured
by a specific tag
Recall Coverage refers to how thoroughly the
topics or ideas within a given tag are captured
Sentiment The expressed feeling or atti-
tude behind a customerrsquos feedback cate-
gorized as positive negative or neutral
Sentiment Phrase Also referred to as a
Sentiment Bearing Phrase or SBP A phrase
or sentence identified with positive nega-
tive or neutral sentiment
Sentiment Score A measure for both the
polarity and intensity of the sentiment
within a given comment
Tag A label generated from text analytics
that groups together similar customer com-
ments around a specific concept or topic
Text Analytics The methods amp pro-
cesses used for obtaining insights from
unstructured data
Text Analytics Model A natural language
processing engine that uses tags to label
and organize unstructured data
Theme A dynamically extracted concept
from a collection of comments generated by
an unsupervised machine learning algorithm
Unstructured Data Qualitative data or
information that is not organized according
to an easily recognizable structure in-
cludes comments social data images or
audio recordings
04
05
05
LET rsquoS TALK ABOUT ACCURACY
What It Is and What It IsnrsquotNow that wersquove gone over the top terms itrsquos time to recognize a truism
vital to understanding text analytics as a whole
06
When we talk about analyzing unstructured data what wersquore really
asking computers to do is understand and analyze a very human
conceptmdashand then to do it in multiple languages
The truth is computers will never be able to understand 100 percent
of the nuances of human language (especially when you consider that
even humans often have trouble distinguishing and understanding
emotion sarcasm slang and all the other complexities inherent in
everyday speech) That is why it is so important to dispel common
misconceptions about accuracy Wersquove laid out a few fictions below
and replaced them with fact
The First Step to Unraveling the Mystery of Text Analytics Is to Understand That It Is Both an Art and a Science
07
DIFFERENTIATING ACCURACY FACT FROM FICTION
Fiction 1
ACCURACY HAS A SINGUL AR MEANING
The term ldquoaccuracyrdquo gets thrown around in text analytics
discussions a lot But what does it actually mean When
referring to the accuracy of the text analytics as a whole
accuracy is based on two factors recall and precision Often
CX vendors will tout 80-90 percent accuracy when what they
are really talking about is precision However what they do not
account for is recallmdashthe extent to which the model retrieves all
key terms or topics For example if the recall rate is only 60
you are missing 40 of actionable intelligence without even
realizing it
In addition when speaking to translation accuracy is often
confused with literalness or the degree of similarity between
exact words and grammar Accuracy is actually the degree of
similarity in meaning A comment does not need to be translated
literally to be accurate
False Negatives
TrueNegatives
Relevant Instances
Retr
ieve
d In
stan
ces
FalsePositives
TruePositives
PRECISION Fraction of retrieved instanced
that are relevant (how useful the results are)
RECALL Fraction of relevant instanced that
are retrieved (how complete the results are)
08
Fiction 3
ACCURACY IS THE GOAL
This may be the biggest misconception alive in text analytics
today Accuracy is not the goal of text analytics itrsquos a happy
byproduct The goal of text analytics in an experience program
is the ability to transform customer data into business
improvement Itrsquos important to never lose sight of that end goal
when designing your program and the text analytics solution that
helps to power it
Fiction 2
PERCENTAGES ARE A TRUE REPRESENTATION
OF ACCURACY
Yoursquove heard accuracy rates thrown around ldquoOur text analytics
are 85 accuraterdquo But have you ever thought about what that
actually means especially when yoursquore digging through your
data When you come across a mis-tagged or miscategorized
comment you might be tempted to think ldquoWell these text
analytics donrsquot workrdquo If you stop and think about it however if
the text analytics are 85 accurate that still means that 15
wonrsquot be And thatrsquos when percentages donrsquot always feel like
a true representation of accuracy Your brain might know that
15 of one thousand comments might mean 150 comments are
tagged incorrectly but when you see it in your data it can feel
like a lot more
08
09
09
TOP OBJECTIVES FOR A MODERN TEXT ANALY TICS SOLUTION
Scalability Quality Actionability and SpeedSo if 100 accuracy is neither achievable nor the end-all-be-all of text analytics solutions what
should experience programs strive for Here are four key areas to consider when measuring the
effectiveness of a text analytics solution
010
Scalability
Any and every growing business (especially those with
a global reach) needs a solution that supports all of the
countries and languages your customers work and buy
inmdashat an acceptable level of quality and at a price that
makes sense No one has unlimited resources so the
first criteria must be that the text analytics solution you
choose is scalable
Quality
The goal of text analytics is not to allow you to read
every single comment that your customers submit
For most businesses this is simply not feasible unless
you have a team of analysts dedicated to it full time
Your text analytics solution must be able to understand
and surface important trends and patterns based on
individual comments and the sentiments behind them
011
Actionability
Another key objective of text analytics is to be able to
take action on the customer issues that most affect your
business This means you need a layer of sophisticated
analytics that can add tags and themes on a granular level
uncover sentiment assign categories identify intent spot
legal issues and possible customer churn and more
Speed
Business today moves quickly so you need a text analytics
solution that can keep pacemdashwhich often means real-time
analysis and reporting This is specifically relevant when
considering translations for global companies The need
for speed means that brands donrsquot have time to wait for a
perfect translation of each and every comment because
perfect translations are only possible if a team manually
translates every word which takes a significant and
unrealistic amount of time amp money
Scalability
Any and every growing business (especially those with
a global reach) needs a solution that supports all of the
countries and languages your customers work and buy
inmdashat an acceptable level of quality and at a price that
makes sense No one has unlimited resources so the
first criteria must be that the text analytics solution you
choose is scalable
012
Tiffany amp Co Leverages Global Text Analytics for Market-Specific and Company-Wide Understanding of Experiences
For a global company the ability to collect and analyze customer feedback across multiple regions and languagesmdash
within a single platformmdashis essential Tiffany amp Co one of the worldrsquos most recognizable and respected luxury
brands leverages InMoment in 18 languages across 300 company-owned stores multiple authorized retailers
its e-commerce platform and in its customer care center for a real-time understanding of market-specific and
company-wide feedback trends
After translation to English comments are analyzed using approximately 150 custom tags and customer sentiment
is displayed in a multitude of reports and dashboards in both English and the original comment language This
way both local and corporate program owners can view results in the language that is most comfortable for them
Further InMomentrsquos highly accurate translation means Tiffany amp Co may use any and every platform featuremdashon
feedback collected in all languagesmdashto sift through comments and research topics most important to the brand
With more than 15 million pieces of customer feedback collected each year comments are used to recognize
exceptional locations and employees identify areas of opportunity and understand the specific drivers of customer
satisfaction The ability to analyze feedback across 18 different languages helps Tiffany amp Co understand nuances
in the experiences and expectations of its customers across each region market and location
013
13
GET TING GLOBAL WITH TRANSL ATION
Human Machine and MoreAs we mentioned in the last section the translation method you choose
can affect the goals you set for the scalability quality actionability and
speed of your text analytics solution
014
Human TranslationHuman translation is the method of having each comment translated
individually as it is submitted by the customer This type of manual
translation requires a long expensive process
Instead of real-time results there is usually a wait of 24 to 48 hours (or
more) to get the translated results And because human translation is
limited to languages that a specific translator can work with outsourcing
more complex languages to other translators will also add additional
costsmdashand more time
LETrsquoS GO OVER THE DIFFERENT OPTIONS
AND THEIR PROS AND CONS
015
Native Language LibrariesA native language library is a text analytics model created
entirely in a language of choice so the system analyzes and
tags comments in that native language instead of translating
them to English tagging them and exporting them back to the
original language Creating a native language library involves a
native speaker building out the initial text analytics model and
then constantly refining it to ensure that it remains up to date
Creating language libraries is a slow and expensive process
usually costing hundreds of thousands of dollars and taking
months or years to build Large amounts of data must be
collected collated and categorized in the language of choice
before the libraries can be built In addition each of these
libraries must be built and tuned to individual industries to
ensure their accuracy The main problem with these native
language libraries therefore is the lack of scalability
016
Machine Translated Human LedMachine translation is exactly what it sounds like It uses
machinesmdashtrained and monitored by humansmdashto translate
text When translating machine translation services look for
patterns and trends in millions of documents to provide the
best translation
Additionally many machine translation services are now using
a neural machine translation engine This means that rather
than translating individual words it uses a neural network
capable of deep learning to look at whole sentences and
understand the greater context These translation services are
highly accurate actionable cost efficient and scalable which
makes them the best option for an international VoC program
017
17
A ldquoMODELrdquo TEXT ANALY TICS SOLUTION
Tailoring Intelligence for Your BusinessThe next element of a text analytics solution to explore is the models you can employ to label
and organize your unstructured data These models have truly come a long way when it comes to their ability to deliver intelligence that is specifically tailored to your business so you can identify
areas that require action and deliver direct value to your bottom line
Traditionally brands were faced with the choice between two specific types of text analytics models industry models and custom models The definition of these two historical approaches
is self explanatory but letrsquos take a deeper dive
018
Industry ModelsAn industry model is developed by experts to reflect the
characteristics of a specific industry These models are
often designed by a vendor to be employed for every client
that fits that particular industry and therefore it often
misses brand-specific insights Additionally because the
same model is used for multiple brands the model canrsquot
be adapted for one brand without being adapted for all
However these models are a much more time-and-cost
effective solution when compared to custom models
Custom ModelsA custom model is built specifically for a certain company
gaining individualized context on the moments that set
a brand apart from its competitors While they are highly
accurate these models are more expensive to create and
maintain so they arenrsquot updated regularly Additionally they
require a high number of comments to test
019
Custom Layered ModelsThere are pros and cons to both of these historical
approaches but there is a new more effective
approach on the scene that gives experience
programs the best of both worlds custom layered
models Custom layered models take the accuracy of
custom models and the efficiency of industry models
to provide tailored accurate and effective results at a
fraction of the cost
How InMomentrsquos Custom Layered Models leverage
over 50 industry models to craft a personalized
tailored solution for each brand And when your text
analytics are built to suit your specific needs the
intelligence you gain from them is multiplied Take a
look at the difference
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retailecommerce
pharmacycustom tags
3xMore Data Points
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retail
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
SINGLE INDUSTRY MODEL CUSTOM L AYERED MODEL
019
020
Comparing and Contrasting ModelsWant a straightforward look at how historical industry and custom models compare to custom layed models
Wersquove laid out the features and benefits of each in this chart
INDUSTRY MODELS CUSTOM MODELS CUSTOM LAYERED MODELS
Covers All Industries
Reflects Unique Business
Delivers Insights for a Combination of Offerings Channels amp Services
Can Be Adapted for an Individual Business without Changing the Model for All Clients Using That Model
Highly Accurate
Cost Effective
Time Efficient
Updated Regularly
Can Be Tested with Limited Data
020
021
21
FACTORING IN SENTIMENT
Getting to the WhyWhen humans look to tell effective storiesmdashwhether in film or in conversationmdashsentiment is
absolutely essential Itrsquos no different when your customer or employee is trying to tell you a
story about their experience via unstructured feedback Itrsquos your text analytics solutionrsquos job to
detect the emotion in their story identify it and help you understand what is causing that specific
sentiment so you can improve experiences But the specific method you use to help your solution
detect sentiment can affect the results you get and therefore your ability to act and transform
022
Rules-Based SentimentTraditionally the way text analytics identified senti-
ment was through a rules-based approach As the name
suggests this method applied rigid rules to understand
unstructured feedback Additionally changing these
rules or updating them required manual intervention
And even with those manual updates accuracy never
really improves
Adaptive Sentiment EngineThe idea that inspired the adaptive sentiment engine
approach is that rigid rules do not lend themselves to best
understand human emotions Instead adaptive sentiment
is AI driven meaning that it learns and improves over time
giving you more accurate results This engine continually
recognizes new terms and phrases as well meaning that
it grows and adapts in lockstep with your brand Check out
the difference below
023
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive Positive Negative NegativeSentiment Score 92
Confidence 98Sentiment Score 52
Confidence 97Sentiment Score -49
Confidence 41Sentiment Score -79
Confidence 94
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
RULES -BASED MACHINE LEARNING
NegativeNeutral
024
Now that you know what it takes to make your text analytics
solution truly world-class what next Well itrsquos time to make
sure that your greater experience program is set up to actually
drive improvement for your organization
Text analytics are just one piece of the puzzle when it comes
to using your VoC VoE or other experience initiatives to
power business growth When you look at the bigger picture
it becomes clear that you need a solid strategy for success
At InMoment we leverage the Continuous Improvement
Framework a method that helps us guide our clients through
designing listening understanding transforming and realizing
their goals for their experience initiative
If you want to learn more about this framework
and how text analytics fit into it you can read
more here
Like what yoursquove seen so far
Here are a few other pieces you may like
How You Listen Matters Modernizing Your Methods amp Approach to Customer Feedback
How to Prove the Business Value of Your CX Program
Building a Business Case for Your CFO (or Anyone Else)
Measuring Doesnrsquot Make You Taller
Therersquos more where that came from on
wwwinmomentcomresources
DRIVING IMPROVEMENT
024
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C
05
05
LET rsquoS TALK ABOUT ACCURACY
What It Is and What It IsnrsquotNow that wersquove gone over the top terms itrsquos time to recognize a truism
vital to understanding text analytics as a whole
06
When we talk about analyzing unstructured data what wersquore really
asking computers to do is understand and analyze a very human
conceptmdashand then to do it in multiple languages
The truth is computers will never be able to understand 100 percent
of the nuances of human language (especially when you consider that
even humans often have trouble distinguishing and understanding
emotion sarcasm slang and all the other complexities inherent in
everyday speech) That is why it is so important to dispel common
misconceptions about accuracy Wersquove laid out a few fictions below
and replaced them with fact
The First Step to Unraveling the Mystery of Text Analytics Is to Understand That It Is Both an Art and a Science
07
DIFFERENTIATING ACCURACY FACT FROM FICTION
Fiction 1
ACCURACY HAS A SINGUL AR MEANING
The term ldquoaccuracyrdquo gets thrown around in text analytics
discussions a lot But what does it actually mean When
referring to the accuracy of the text analytics as a whole
accuracy is based on two factors recall and precision Often
CX vendors will tout 80-90 percent accuracy when what they
are really talking about is precision However what they do not
account for is recallmdashthe extent to which the model retrieves all
key terms or topics For example if the recall rate is only 60
you are missing 40 of actionable intelligence without even
realizing it
In addition when speaking to translation accuracy is often
confused with literalness or the degree of similarity between
exact words and grammar Accuracy is actually the degree of
similarity in meaning A comment does not need to be translated
literally to be accurate
False Negatives
TrueNegatives
Relevant Instances
Retr
ieve
d In
stan
ces
FalsePositives
TruePositives
PRECISION Fraction of retrieved instanced
that are relevant (how useful the results are)
RECALL Fraction of relevant instanced that
are retrieved (how complete the results are)
08
Fiction 3
ACCURACY IS THE GOAL
This may be the biggest misconception alive in text analytics
today Accuracy is not the goal of text analytics itrsquos a happy
byproduct The goal of text analytics in an experience program
is the ability to transform customer data into business
improvement Itrsquos important to never lose sight of that end goal
when designing your program and the text analytics solution that
helps to power it
Fiction 2
PERCENTAGES ARE A TRUE REPRESENTATION
OF ACCURACY
Yoursquove heard accuracy rates thrown around ldquoOur text analytics
are 85 accuraterdquo But have you ever thought about what that
actually means especially when yoursquore digging through your
data When you come across a mis-tagged or miscategorized
comment you might be tempted to think ldquoWell these text
analytics donrsquot workrdquo If you stop and think about it however if
the text analytics are 85 accurate that still means that 15
wonrsquot be And thatrsquos when percentages donrsquot always feel like
a true representation of accuracy Your brain might know that
15 of one thousand comments might mean 150 comments are
tagged incorrectly but when you see it in your data it can feel
like a lot more
08
09
09
TOP OBJECTIVES FOR A MODERN TEXT ANALY TICS SOLUTION
Scalability Quality Actionability and SpeedSo if 100 accuracy is neither achievable nor the end-all-be-all of text analytics solutions what
should experience programs strive for Here are four key areas to consider when measuring the
effectiveness of a text analytics solution
010
Scalability
Any and every growing business (especially those with
a global reach) needs a solution that supports all of the
countries and languages your customers work and buy
inmdashat an acceptable level of quality and at a price that
makes sense No one has unlimited resources so the
first criteria must be that the text analytics solution you
choose is scalable
Quality
The goal of text analytics is not to allow you to read
every single comment that your customers submit
For most businesses this is simply not feasible unless
you have a team of analysts dedicated to it full time
Your text analytics solution must be able to understand
and surface important trends and patterns based on
individual comments and the sentiments behind them
011
Actionability
Another key objective of text analytics is to be able to
take action on the customer issues that most affect your
business This means you need a layer of sophisticated
analytics that can add tags and themes on a granular level
uncover sentiment assign categories identify intent spot
legal issues and possible customer churn and more
Speed
Business today moves quickly so you need a text analytics
solution that can keep pacemdashwhich often means real-time
analysis and reporting This is specifically relevant when
considering translations for global companies The need
for speed means that brands donrsquot have time to wait for a
perfect translation of each and every comment because
perfect translations are only possible if a team manually
translates every word which takes a significant and
unrealistic amount of time amp money
Scalability
Any and every growing business (especially those with
a global reach) needs a solution that supports all of the
countries and languages your customers work and buy
inmdashat an acceptable level of quality and at a price that
makes sense No one has unlimited resources so the
first criteria must be that the text analytics solution you
choose is scalable
012
Tiffany amp Co Leverages Global Text Analytics for Market-Specific and Company-Wide Understanding of Experiences
For a global company the ability to collect and analyze customer feedback across multiple regions and languagesmdash
within a single platformmdashis essential Tiffany amp Co one of the worldrsquos most recognizable and respected luxury
brands leverages InMoment in 18 languages across 300 company-owned stores multiple authorized retailers
its e-commerce platform and in its customer care center for a real-time understanding of market-specific and
company-wide feedback trends
After translation to English comments are analyzed using approximately 150 custom tags and customer sentiment
is displayed in a multitude of reports and dashboards in both English and the original comment language This
way both local and corporate program owners can view results in the language that is most comfortable for them
Further InMomentrsquos highly accurate translation means Tiffany amp Co may use any and every platform featuremdashon
feedback collected in all languagesmdashto sift through comments and research topics most important to the brand
With more than 15 million pieces of customer feedback collected each year comments are used to recognize
exceptional locations and employees identify areas of opportunity and understand the specific drivers of customer
satisfaction The ability to analyze feedback across 18 different languages helps Tiffany amp Co understand nuances
in the experiences and expectations of its customers across each region market and location
013
13
GET TING GLOBAL WITH TRANSL ATION
Human Machine and MoreAs we mentioned in the last section the translation method you choose
can affect the goals you set for the scalability quality actionability and
speed of your text analytics solution
014
Human TranslationHuman translation is the method of having each comment translated
individually as it is submitted by the customer This type of manual
translation requires a long expensive process
Instead of real-time results there is usually a wait of 24 to 48 hours (or
more) to get the translated results And because human translation is
limited to languages that a specific translator can work with outsourcing
more complex languages to other translators will also add additional
costsmdashand more time
LETrsquoS GO OVER THE DIFFERENT OPTIONS
AND THEIR PROS AND CONS
015
Native Language LibrariesA native language library is a text analytics model created
entirely in a language of choice so the system analyzes and
tags comments in that native language instead of translating
them to English tagging them and exporting them back to the
original language Creating a native language library involves a
native speaker building out the initial text analytics model and
then constantly refining it to ensure that it remains up to date
Creating language libraries is a slow and expensive process
usually costing hundreds of thousands of dollars and taking
months or years to build Large amounts of data must be
collected collated and categorized in the language of choice
before the libraries can be built In addition each of these
libraries must be built and tuned to individual industries to
ensure their accuracy The main problem with these native
language libraries therefore is the lack of scalability
016
Machine Translated Human LedMachine translation is exactly what it sounds like It uses
machinesmdashtrained and monitored by humansmdashto translate
text When translating machine translation services look for
patterns and trends in millions of documents to provide the
best translation
Additionally many machine translation services are now using
a neural machine translation engine This means that rather
than translating individual words it uses a neural network
capable of deep learning to look at whole sentences and
understand the greater context These translation services are
highly accurate actionable cost efficient and scalable which
makes them the best option for an international VoC program
017
17
A ldquoMODELrdquo TEXT ANALY TICS SOLUTION
Tailoring Intelligence for Your BusinessThe next element of a text analytics solution to explore is the models you can employ to label
and organize your unstructured data These models have truly come a long way when it comes to their ability to deliver intelligence that is specifically tailored to your business so you can identify
areas that require action and deliver direct value to your bottom line
Traditionally brands were faced with the choice between two specific types of text analytics models industry models and custom models The definition of these two historical approaches
is self explanatory but letrsquos take a deeper dive
018
Industry ModelsAn industry model is developed by experts to reflect the
characteristics of a specific industry These models are
often designed by a vendor to be employed for every client
that fits that particular industry and therefore it often
misses brand-specific insights Additionally because the
same model is used for multiple brands the model canrsquot
be adapted for one brand without being adapted for all
However these models are a much more time-and-cost
effective solution when compared to custom models
Custom ModelsA custom model is built specifically for a certain company
gaining individualized context on the moments that set
a brand apart from its competitors While they are highly
accurate these models are more expensive to create and
maintain so they arenrsquot updated regularly Additionally they
require a high number of comments to test
019
Custom Layered ModelsThere are pros and cons to both of these historical
approaches but there is a new more effective
approach on the scene that gives experience
programs the best of both worlds custom layered
models Custom layered models take the accuracy of
custom models and the efficiency of industry models
to provide tailored accurate and effective results at a
fraction of the cost
How InMomentrsquos Custom Layered Models leverage
over 50 industry models to craft a personalized
tailored solution for each brand And when your text
analytics are built to suit your specific needs the
intelligence you gain from them is multiplied Take a
look at the difference
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retailecommerce
pharmacycustom tags
3xMore Data Points
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retail
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
SINGLE INDUSTRY MODEL CUSTOM L AYERED MODEL
019
020
Comparing and Contrasting ModelsWant a straightforward look at how historical industry and custom models compare to custom layed models
Wersquove laid out the features and benefits of each in this chart
INDUSTRY MODELS CUSTOM MODELS CUSTOM LAYERED MODELS
Covers All Industries
Reflects Unique Business
Delivers Insights for a Combination of Offerings Channels amp Services
Can Be Adapted for an Individual Business without Changing the Model for All Clients Using That Model
Highly Accurate
Cost Effective
Time Efficient
Updated Regularly
Can Be Tested with Limited Data
020
021
21
FACTORING IN SENTIMENT
Getting to the WhyWhen humans look to tell effective storiesmdashwhether in film or in conversationmdashsentiment is
absolutely essential Itrsquos no different when your customer or employee is trying to tell you a
story about their experience via unstructured feedback Itrsquos your text analytics solutionrsquos job to
detect the emotion in their story identify it and help you understand what is causing that specific
sentiment so you can improve experiences But the specific method you use to help your solution
detect sentiment can affect the results you get and therefore your ability to act and transform
022
Rules-Based SentimentTraditionally the way text analytics identified senti-
ment was through a rules-based approach As the name
suggests this method applied rigid rules to understand
unstructured feedback Additionally changing these
rules or updating them required manual intervention
And even with those manual updates accuracy never
really improves
Adaptive Sentiment EngineThe idea that inspired the adaptive sentiment engine
approach is that rigid rules do not lend themselves to best
understand human emotions Instead adaptive sentiment
is AI driven meaning that it learns and improves over time
giving you more accurate results This engine continually
recognizes new terms and phrases as well meaning that
it grows and adapts in lockstep with your brand Check out
the difference below
023
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive Positive Negative NegativeSentiment Score 92
Confidence 98Sentiment Score 52
Confidence 97Sentiment Score -49
Confidence 41Sentiment Score -79
Confidence 94
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
RULES -BASED MACHINE LEARNING
NegativeNeutral
024
Now that you know what it takes to make your text analytics
solution truly world-class what next Well itrsquos time to make
sure that your greater experience program is set up to actually
drive improvement for your organization
Text analytics are just one piece of the puzzle when it comes
to using your VoC VoE or other experience initiatives to
power business growth When you look at the bigger picture
it becomes clear that you need a solid strategy for success
At InMoment we leverage the Continuous Improvement
Framework a method that helps us guide our clients through
designing listening understanding transforming and realizing
their goals for their experience initiative
If you want to learn more about this framework
and how text analytics fit into it you can read
more here
Like what yoursquove seen so far
Here are a few other pieces you may like
How You Listen Matters Modernizing Your Methods amp Approach to Customer Feedback
How to Prove the Business Value of Your CX Program
Building a Business Case for Your CFO (or Anyone Else)
Measuring Doesnrsquot Make You Taller
Therersquos more where that came from on
wwwinmomentcomresources
DRIVING IMPROVEMENT
024
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C
06
When we talk about analyzing unstructured data what wersquore really
asking computers to do is understand and analyze a very human
conceptmdashand then to do it in multiple languages
The truth is computers will never be able to understand 100 percent
of the nuances of human language (especially when you consider that
even humans often have trouble distinguishing and understanding
emotion sarcasm slang and all the other complexities inherent in
everyday speech) That is why it is so important to dispel common
misconceptions about accuracy Wersquove laid out a few fictions below
and replaced them with fact
The First Step to Unraveling the Mystery of Text Analytics Is to Understand That It Is Both an Art and a Science
07
DIFFERENTIATING ACCURACY FACT FROM FICTION
Fiction 1
ACCURACY HAS A SINGUL AR MEANING
The term ldquoaccuracyrdquo gets thrown around in text analytics
discussions a lot But what does it actually mean When
referring to the accuracy of the text analytics as a whole
accuracy is based on two factors recall and precision Often
CX vendors will tout 80-90 percent accuracy when what they
are really talking about is precision However what they do not
account for is recallmdashthe extent to which the model retrieves all
key terms or topics For example if the recall rate is only 60
you are missing 40 of actionable intelligence without even
realizing it
In addition when speaking to translation accuracy is often
confused with literalness or the degree of similarity between
exact words and grammar Accuracy is actually the degree of
similarity in meaning A comment does not need to be translated
literally to be accurate
False Negatives
TrueNegatives
Relevant Instances
Retr
ieve
d In
stan
ces
FalsePositives
TruePositives
PRECISION Fraction of retrieved instanced
that are relevant (how useful the results are)
RECALL Fraction of relevant instanced that
are retrieved (how complete the results are)
08
Fiction 3
ACCURACY IS THE GOAL
This may be the biggest misconception alive in text analytics
today Accuracy is not the goal of text analytics itrsquos a happy
byproduct The goal of text analytics in an experience program
is the ability to transform customer data into business
improvement Itrsquos important to never lose sight of that end goal
when designing your program and the text analytics solution that
helps to power it
Fiction 2
PERCENTAGES ARE A TRUE REPRESENTATION
OF ACCURACY
Yoursquove heard accuracy rates thrown around ldquoOur text analytics
are 85 accuraterdquo But have you ever thought about what that
actually means especially when yoursquore digging through your
data When you come across a mis-tagged or miscategorized
comment you might be tempted to think ldquoWell these text
analytics donrsquot workrdquo If you stop and think about it however if
the text analytics are 85 accurate that still means that 15
wonrsquot be And thatrsquos when percentages donrsquot always feel like
a true representation of accuracy Your brain might know that
15 of one thousand comments might mean 150 comments are
tagged incorrectly but when you see it in your data it can feel
like a lot more
08
09
09
TOP OBJECTIVES FOR A MODERN TEXT ANALY TICS SOLUTION
Scalability Quality Actionability and SpeedSo if 100 accuracy is neither achievable nor the end-all-be-all of text analytics solutions what
should experience programs strive for Here are four key areas to consider when measuring the
effectiveness of a text analytics solution
010
Scalability
Any and every growing business (especially those with
a global reach) needs a solution that supports all of the
countries and languages your customers work and buy
inmdashat an acceptable level of quality and at a price that
makes sense No one has unlimited resources so the
first criteria must be that the text analytics solution you
choose is scalable
Quality
The goal of text analytics is not to allow you to read
every single comment that your customers submit
For most businesses this is simply not feasible unless
you have a team of analysts dedicated to it full time
Your text analytics solution must be able to understand
and surface important trends and patterns based on
individual comments and the sentiments behind them
011
Actionability
Another key objective of text analytics is to be able to
take action on the customer issues that most affect your
business This means you need a layer of sophisticated
analytics that can add tags and themes on a granular level
uncover sentiment assign categories identify intent spot
legal issues and possible customer churn and more
Speed
Business today moves quickly so you need a text analytics
solution that can keep pacemdashwhich often means real-time
analysis and reporting This is specifically relevant when
considering translations for global companies The need
for speed means that brands donrsquot have time to wait for a
perfect translation of each and every comment because
perfect translations are only possible if a team manually
translates every word which takes a significant and
unrealistic amount of time amp money
Scalability
Any and every growing business (especially those with
a global reach) needs a solution that supports all of the
countries and languages your customers work and buy
inmdashat an acceptable level of quality and at a price that
makes sense No one has unlimited resources so the
first criteria must be that the text analytics solution you
choose is scalable
012
Tiffany amp Co Leverages Global Text Analytics for Market-Specific and Company-Wide Understanding of Experiences
For a global company the ability to collect and analyze customer feedback across multiple regions and languagesmdash
within a single platformmdashis essential Tiffany amp Co one of the worldrsquos most recognizable and respected luxury
brands leverages InMoment in 18 languages across 300 company-owned stores multiple authorized retailers
its e-commerce platform and in its customer care center for a real-time understanding of market-specific and
company-wide feedback trends
After translation to English comments are analyzed using approximately 150 custom tags and customer sentiment
is displayed in a multitude of reports and dashboards in both English and the original comment language This
way both local and corporate program owners can view results in the language that is most comfortable for them
Further InMomentrsquos highly accurate translation means Tiffany amp Co may use any and every platform featuremdashon
feedback collected in all languagesmdashto sift through comments and research topics most important to the brand
With more than 15 million pieces of customer feedback collected each year comments are used to recognize
exceptional locations and employees identify areas of opportunity and understand the specific drivers of customer
satisfaction The ability to analyze feedback across 18 different languages helps Tiffany amp Co understand nuances
in the experiences and expectations of its customers across each region market and location
013
13
GET TING GLOBAL WITH TRANSL ATION
Human Machine and MoreAs we mentioned in the last section the translation method you choose
can affect the goals you set for the scalability quality actionability and
speed of your text analytics solution
014
Human TranslationHuman translation is the method of having each comment translated
individually as it is submitted by the customer This type of manual
translation requires a long expensive process
Instead of real-time results there is usually a wait of 24 to 48 hours (or
more) to get the translated results And because human translation is
limited to languages that a specific translator can work with outsourcing
more complex languages to other translators will also add additional
costsmdashand more time
LETrsquoS GO OVER THE DIFFERENT OPTIONS
AND THEIR PROS AND CONS
015
Native Language LibrariesA native language library is a text analytics model created
entirely in a language of choice so the system analyzes and
tags comments in that native language instead of translating
them to English tagging them and exporting them back to the
original language Creating a native language library involves a
native speaker building out the initial text analytics model and
then constantly refining it to ensure that it remains up to date
Creating language libraries is a slow and expensive process
usually costing hundreds of thousands of dollars and taking
months or years to build Large amounts of data must be
collected collated and categorized in the language of choice
before the libraries can be built In addition each of these
libraries must be built and tuned to individual industries to
ensure their accuracy The main problem with these native
language libraries therefore is the lack of scalability
016
Machine Translated Human LedMachine translation is exactly what it sounds like It uses
machinesmdashtrained and monitored by humansmdashto translate
text When translating machine translation services look for
patterns and trends in millions of documents to provide the
best translation
Additionally many machine translation services are now using
a neural machine translation engine This means that rather
than translating individual words it uses a neural network
capable of deep learning to look at whole sentences and
understand the greater context These translation services are
highly accurate actionable cost efficient and scalable which
makes them the best option for an international VoC program
017
17
A ldquoMODELrdquo TEXT ANALY TICS SOLUTION
Tailoring Intelligence for Your BusinessThe next element of a text analytics solution to explore is the models you can employ to label
and organize your unstructured data These models have truly come a long way when it comes to their ability to deliver intelligence that is specifically tailored to your business so you can identify
areas that require action and deliver direct value to your bottom line
Traditionally brands were faced with the choice between two specific types of text analytics models industry models and custom models The definition of these two historical approaches
is self explanatory but letrsquos take a deeper dive
018
Industry ModelsAn industry model is developed by experts to reflect the
characteristics of a specific industry These models are
often designed by a vendor to be employed for every client
that fits that particular industry and therefore it often
misses brand-specific insights Additionally because the
same model is used for multiple brands the model canrsquot
be adapted for one brand without being adapted for all
However these models are a much more time-and-cost
effective solution when compared to custom models
Custom ModelsA custom model is built specifically for a certain company
gaining individualized context on the moments that set
a brand apart from its competitors While they are highly
accurate these models are more expensive to create and
maintain so they arenrsquot updated regularly Additionally they
require a high number of comments to test
019
Custom Layered ModelsThere are pros and cons to both of these historical
approaches but there is a new more effective
approach on the scene that gives experience
programs the best of both worlds custom layered
models Custom layered models take the accuracy of
custom models and the efficiency of industry models
to provide tailored accurate and effective results at a
fraction of the cost
How InMomentrsquos Custom Layered Models leverage
over 50 industry models to craft a personalized
tailored solution for each brand And when your text
analytics are built to suit your specific needs the
intelligence you gain from them is multiplied Take a
look at the difference
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retailecommerce
pharmacycustom tags
3xMore Data Points
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retail
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
SINGLE INDUSTRY MODEL CUSTOM L AYERED MODEL
019
020
Comparing and Contrasting ModelsWant a straightforward look at how historical industry and custom models compare to custom layed models
Wersquove laid out the features and benefits of each in this chart
INDUSTRY MODELS CUSTOM MODELS CUSTOM LAYERED MODELS
Covers All Industries
Reflects Unique Business
Delivers Insights for a Combination of Offerings Channels amp Services
Can Be Adapted for an Individual Business without Changing the Model for All Clients Using That Model
Highly Accurate
Cost Effective
Time Efficient
Updated Regularly
Can Be Tested with Limited Data
020
021
21
FACTORING IN SENTIMENT
Getting to the WhyWhen humans look to tell effective storiesmdashwhether in film or in conversationmdashsentiment is
absolutely essential Itrsquos no different when your customer or employee is trying to tell you a
story about their experience via unstructured feedback Itrsquos your text analytics solutionrsquos job to
detect the emotion in their story identify it and help you understand what is causing that specific
sentiment so you can improve experiences But the specific method you use to help your solution
detect sentiment can affect the results you get and therefore your ability to act and transform
022
Rules-Based SentimentTraditionally the way text analytics identified senti-
ment was through a rules-based approach As the name
suggests this method applied rigid rules to understand
unstructured feedback Additionally changing these
rules or updating them required manual intervention
And even with those manual updates accuracy never
really improves
Adaptive Sentiment EngineThe idea that inspired the adaptive sentiment engine
approach is that rigid rules do not lend themselves to best
understand human emotions Instead adaptive sentiment
is AI driven meaning that it learns and improves over time
giving you more accurate results This engine continually
recognizes new terms and phrases as well meaning that
it grows and adapts in lockstep with your brand Check out
the difference below
023
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive Positive Negative NegativeSentiment Score 92
Confidence 98Sentiment Score 52
Confidence 97Sentiment Score -49
Confidence 41Sentiment Score -79
Confidence 94
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
RULES -BASED MACHINE LEARNING
NegativeNeutral
024
Now that you know what it takes to make your text analytics
solution truly world-class what next Well itrsquos time to make
sure that your greater experience program is set up to actually
drive improvement for your organization
Text analytics are just one piece of the puzzle when it comes
to using your VoC VoE or other experience initiatives to
power business growth When you look at the bigger picture
it becomes clear that you need a solid strategy for success
At InMoment we leverage the Continuous Improvement
Framework a method that helps us guide our clients through
designing listening understanding transforming and realizing
their goals for their experience initiative
If you want to learn more about this framework
and how text analytics fit into it you can read
more here
Like what yoursquove seen so far
Here are a few other pieces you may like
How You Listen Matters Modernizing Your Methods amp Approach to Customer Feedback
How to Prove the Business Value of Your CX Program
Building a Business Case for Your CFO (or Anyone Else)
Measuring Doesnrsquot Make You Taller
Therersquos more where that came from on
wwwinmomentcomresources
DRIVING IMPROVEMENT
024
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C
07
DIFFERENTIATING ACCURACY FACT FROM FICTION
Fiction 1
ACCURACY HAS A SINGUL AR MEANING
The term ldquoaccuracyrdquo gets thrown around in text analytics
discussions a lot But what does it actually mean When
referring to the accuracy of the text analytics as a whole
accuracy is based on two factors recall and precision Often
CX vendors will tout 80-90 percent accuracy when what they
are really talking about is precision However what they do not
account for is recallmdashthe extent to which the model retrieves all
key terms or topics For example if the recall rate is only 60
you are missing 40 of actionable intelligence without even
realizing it
In addition when speaking to translation accuracy is often
confused with literalness or the degree of similarity between
exact words and grammar Accuracy is actually the degree of
similarity in meaning A comment does not need to be translated
literally to be accurate
False Negatives
TrueNegatives
Relevant Instances
Retr
ieve
d In
stan
ces
FalsePositives
TruePositives
PRECISION Fraction of retrieved instanced
that are relevant (how useful the results are)
RECALL Fraction of relevant instanced that
are retrieved (how complete the results are)
08
Fiction 3
ACCURACY IS THE GOAL
This may be the biggest misconception alive in text analytics
today Accuracy is not the goal of text analytics itrsquos a happy
byproduct The goal of text analytics in an experience program
is the ability to transform customer data into business
improvement Itrsquos important to never lose sight of that end goal
when designing your program and the text analytics solution that
helps to power it
Fiction 2
PERCENTAGES ARE A TRUE REPRESENTATION
OF ACCURACY
Yoursquove heard accuracy rates thrown around ldquoOur text analytics
are 85 accuraterdquo But have you ever thought about what that
actually means especially when yoursquore digging through your
data When you come across a mis-tagged or miscategorized
comment you might be tempted to think ldquoWell these text
analytics donrsquot workrdquo If you stop and think about it however if
the text analytics are 85 accurate that still means that 15
wonrsquot be And thatrsquos when percentages donrsquot always feel like
a true representation of accuracy Your brain might know that
15 of one thousand comments might mean 150 comments are
tagged incorrectly but when you see it in your data it can feel
like a lot more
08
09
09
TOP OBJECTIVES FOR A MODERN TEXT ANALY TICS SOLUTION
Scalability Quality Actionability and SpeedSo if 100 accuracy is neither achievable nor the end-all-be-all of text analytics solutions what
should experience programs strive for Here are four key areas to consider when measuring the
effectiveness of a text analytics solution
010
Scalability
Any and every growing business (especially those with
a global reach) needs a solution that supports all of the
countries and languages your customers work and buy
inmdashat an acceptable level of quality and at a price that
makes sense No one has unlimited resources so the
first criteria must be that the text analytics solution you
choose is scalable
Quality
The goal of text analytics is not to allow you to read
every single comment that your customers submit
For most businesses this is simply not feasible unless
you have a team of analysts dedicated to it full time
Your text analytics solution must be able to understand
and surface important trends and patterns based on
individual comments and the sentiments behind them
011
Actionability
Another key objective of text analytics is to be able to
take action on the customer issues that most affect your
business This means you need a layer of sophisticated
analytics that can add tags and themes on a granular level
uncover sentiment assign categories identify intent spot
legal issues and possible customer churn and more
Speed
Business today moves quickly so you need a text analytics
solution that can keep pacemdashwhich often means real-time
analysis and reporting This is specifically relevant when
considering translations for global companies The need
for speed means that brands donrsquot have time to wait for a
perfect translation of each and every comment because
perfect translations are only possible if a team manually
translates every word which takes a significant and
unrealistic amount of time amp money
Scalability
Any and every growing business (especially those with
a global reach) needs a solution that supports all of the
countries and languages your customers work and buy
inmdashat an acceptable level of quality and at a price that
makes sense No one has unlimited resources so the
first criteria must be that the text analytics solution you
choose is scalable
012
Tiffany amp Co Leverages Global Text Analytics for Market-Specific and Company-Wide Understanding of Experiences
For a global company the ability to collect and analyze customer feedback across multiple regions and languagesmdash
within a single platformmdashis essential Tiffany amp Co one of the worldrsquos most recognizable and respected luxury
brands leverages InMoment in 18 languages across 300 company-owned stores multiple authorized retailers
its e-commerce platform and in its customer care center for a real-time understanding of market-specific and
company-wide feedback trends
After translation to English comments are analyzed using approximately 150 custom tags and customer sentiment
is displayed in a multitude of reports and dashboards in both English and the original comment language This
way both local and corporate program owners can view results in the language that is most comfortable for them
Further InMomentrsquos highly accurate translation means Tiffany amp Co may use any and every platform featuremdashon
feedback collected in all languagesmdashto sift through comments and research topics most important to the brand
With more than 15 million pieces of customer feedback collected each year comments are used to recognize
exceptional locations and employees identify areas of opportunity and understand the specific drivers of customer
satisfaction The ability to analyze feedback across 18 different languages helps Tiffany amp Co understand nuances
in the experiences and expectations of its customers across each region market and location
013
13
GET TING GLOBAL WITH TRANSL ATION
Human Machine and MoreAs we mentioned in the last section the translation method you choose
can affect the goals you set for the scalability quality actionability and
speed of your text analytics solution
014
Human TranslationHuman translation is the method of having each comment translated
individually as it is submitted by the customer This type of manual
translation requires a long expensive process
Instead of real-time results there is usually a wait of 24 to 48 hours (or
more) to get the translated results And because human translation is
limited to languages that a specific translator can work with outsourcing
more complex languages to other translators will also add additional
costsmdashand more time
LETrsquoS GO OVER THE DIFFERENT OPTIONS
AND THEIR PROS AND CONS
015
Native Language LibrariesA native language library is a text analytics model created
entirely in a language of choice so the system analyzes and
tags comments in that native language instead of translating
them to English tagging them and exporting them back to the
original language Creating a native language library involves a
native speaker building out the initial text analytics model and
then constantly refining it to ensure that it remains up to date
Creating language libraries is a slow and expensive process
usually costing hundreds of thousands of dollars and taking
months or years to build Large amounts of data must be
collected collated and categorized in the language of choice
before the libraries can be built In addition each of these
libraries must be built and tuned to individual industries to
ensure their accuracy The main problem with these native
language libraries therefore is the lack of scalability
016
Machine Translated Human LedMachine translation is exactly what it sounds like It uses
machinesmdashtrained and monitored by humansmdashto translate
text When translating machine translation services look for
patterns and trends in millions of documents to provide the
best translation
Additionally many machine translation services are now using
a neural machine translation engine This means that rather
than translating individual words it uses a neural network
capable of deep learning to look at whole sentences and
understand the greater context These translation services are
highly accurate actionable cost efficient and scalable which
makes them the best option for an international VoC program
017
17
A ldquoMODELrdquo TEXT ANALY TICS SOLUTION
Tailoring Intelligence for Your BusinessThe next element of a text analytics solution to explore is the models you can employ to label
and organize your unstructured data These models have truly come a long way when it comes to their ability to deliver intelligence that is specifically tailored to your business so you can identify
areas that require action and deliver direct value to your bottom line
Traditionally brands were faced with the choice between two specific types of text analytics models industry models and custom models The definition of these two historical approaches
is self explanatory but letrsquos take a deeper dive
018
Industry ModelsAn industry model is developed by experts to reflect the
characteristics of a specific industry These models are
often designed by a vendor to be employed for every client
that fits that particular industry and therefore it often
misses brand-specific insights Additionally because the
same model is used for multiple brands the model canrsquot
be adapted for one brand without being adapted for all
However these models are a much more time-and-cost
effective solution when compared to custom models
Custom ModelsA custom model is built specifically for a certain company
gaining individualized context on the moments that set
a brand apart from its competitors While they are highly
accurate these models are more expensive to create and
maintain so they arenrsquot updated regularly Additionally they
require a high number of comments to test
019
Custom Layered ModelsThere are pros and cons to both of these historical
approaches but there is a new more effective
approach on the scene that gives experience
programs the best of both worlds custom layered
models Custom layered models take the accuracy of
custom models and the efficiency of industry models
to provide tailored accurate and effective results at a
fraction of the cost
How InMomentrsquos Custom Layered Models leverage
over 50 industry models to craft a personalized
tailored solution for each brand And when your text
analytics are built to suit your specific needs the
intelligence you gain from them is multiplied Take a
look at the difference
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retailecommerce
pharmacycustom tags
3xMore Data Points
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retail
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
SINGLE INDUSTRY MODEL CUSTOM L AYERED MODEL
019
020
Comparing and Contrasting ModelsWant a straightforward look at how historical industry and custom models compare to custom layed models
Wersquove laid out the features and benefits of each in this chart
INDUSTRY MODELS CUSTOM MODELS CUSTOM LAYERED MODELS
Covers All Industries
Reflects Unique Business
Delivers Insights for a Combination of Offerings Channels amp Services
Can Be Adapted for an Individual Business without Changing the Model for All Clients Using That Model
Highly Accurate
Cost Effective
Time Efficient
Updated Regularly
Can Be Tested with Limited Data
020
021
21
FACTORING IN SENTIMENT
Getting to the WhyWhen humans look to tell effective storiesmdashwhether in film or in conversationmdashsentiment is
absolutely essential Itrsquos no different when your customer or employee is trying to tell you a
story about their experience via unstructured feedback Itrsquos your text analytics solutionrsquos job to
detect the emotion in their story identify it and help you understand what is causing that specific
sentiment so you can improve experiences But the specific method you use to help your solution
detect sentiment can affect the results you get and therefore your ability to act and transform
022
Rules-Based SentimentTraditionally the way text analytics identified senti-
ment was through a rules-based approach As the name
suggests this method applied rigid rules to understand
unstructured feedback Additionally changing these
rules or updating them required manual intervention
And even with those manual updates accuracy never
really improves
Adaptive Sentiment EngineThe idea that inspired the adaptive sentiment engine
approach is that rigid rules do not lend themselves to best
understand human emotions Instead adaptive sentiment
is AI driven meaning that it learns and improves over time
giving you more accurate results This engine continually
recognizes new terms and phrases as well meaning that
it grows and adapts in lockstep with your brand Check out
the difference below
023
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive Positive Negative NegativeSentiment Score 92
Confidence 98Sentiment Score 52
Confidence 97Sentiment Score -49
Confidence 41Sentiment Score -79
Confidence 94
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
RULES -BASED MACHINE LEARNING
NegativeNeutral
024
Now that you know what it takes to make your text analytics
solution truly world-class what next Well itrsquos time to make
sure that your greater experience program is set up to actually
drive improvement for your organization
Text analytics are just one piece of the puzzle when it comes
to using your VoC VoE or other experience initiatives to
power business growth When you look at the bigger picture
it becomes clear that you need a solid strategy for success
At InMoment we leverage the Continuous Improvement
Framework a method that helps us guide our clients through
designing listening understanding transforming and realizing
their goals for their experience initiative
If you want to learn more about this framework
and how text analytics fit into it you can read
more here
Like what yoursquove seen so far
Here are a few other pieces you may like
How You Listen Matters Modernizing Your Methods amp Approach to Customer Feedback
How to Prove the Business Value of Your CX Program
Building a Business Case for Your CFO (or Anyone Else)
Measuring Doesnrsquot Make You Taller
Therersquos more where that came from on
wwwinmomentcomresources
DRIVING IMPROVEMENT
024
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C
08
Fiction 3
ACCURACY IS THE GOAL
This may be the biggest misconception alive in text analytics
today Accuracy is not the goal of text analytics itrsquos a happy
byproduct The goal of text analytics in an experience program
is the ability to transform customer data into business
improvement Itrsquos important to never lose sight of that end goal
when designing your program and the text analytics solution that
helps to power it
Fiction 2
PERCENTAGES ARE A TRUE REPRESENTATION
OF ACCURACY
Yoursquove heard accuracy rates thrown around ldquoOur text analytics
are 85 accuraterdquo But have you ever thought about what that
actually means especially when yoursquore digging through your
data When you come across a mis-tagged or miscategorized
comment you might be tempted to think ldquoWell these text
analytics donrsquot workrdquo If you stop and think about it however if
the text analytics are 85 accurate that still means that 15
wonrsquot be And thatrsquos when percentages donrsquot always feel like
a true representation of accuracy Your brain might know that
15 of one thousand comments might mean 150 comments are
tagged incorrectly but when you see it in your data it can feel
like a lot more
08
09
09
TOP OBJECTIVES FOR A MODERN TEXT ANALY TICS SOLUTION
Scalability Quality Actionability and SpeedSo if 100 accuracy is neither achievable nor the end-all-be-all of text analytics solutions what
should experience programs strive for Here are four key areas to consider when measuring the
effectiveness of a text analytics solution
010
Scalability
Any and every growing business (especially those with
a global reach) needs a solution that supports all of the
countries and languages your customers work and buy
inmdashat an acceptable level of quality and at a price that
makes sense No one has unlimited resources so the
first criteria must be that the text analytics solution you
choose is scalable
Quality
The goal of text analytics is not to allow you to read
every single comment that your customers submit
For most businesses this is simply not feasible unless
you have a team of analysts dedicated to it full time
Your text analytics solution must be able to understand
and surface important trends and patterns based on
individual comments and the sentiments behind them
011
Actionability
Another key objective of text analytics is to be able to
take action on the customer issues that most affect your
business This means you need a layer of sophisticated
analytics that can add tags and themes on a granular level
uncover sentiment assign categories identify intent spot
legal issues and possible customer churn and more
Speed
Business today moves quickly so you need a text analytics
solution that can keep pacemdashwhich often means real-time
analysis and reporting This is specifically relevant when
considering translations for global companies The need
for speed means that brands donrsquot have time to wait for a
perfect translation of each and every comment because
perfect translations are only possible if a team manually
translates every word which takes a significant and
unrealistic amount of time amp money
Scalability
Any and every growing business (especially those with
a global reach) needs a solution that supports all of the
countries and languages your customers work and buy
inmdashat an acceptable level of quality and at a price that
makes sense No one has unlimited resources so the
first criteria must be that the text analytics solution you
choose is scalable
012
Tiffany amp Co Leverages Global Text Analytics for Market-Specific and Company-Wide Understanding of Experiences
For a global company the ability to collect and analyze customer feedback across multiple regions and languagesmdash
within a single platformmdashis essential Tiffany amp Co one of the worldrsquos most recognizable and respected luxury
brands leverages InMoment in 18 languages across 300 company-owned stores multiple authorized retailers
its e-commerce platform and in its customer care center for a real-time understanding of market-specific and
company-wide feedback trends
After translation to English comments are analyzed using approximately 150 custom tags and customer sentiment
is displayed in a multitude of reports and dashboards in both English and the original comment language This
way both local and corporate program owners can view results in the language that is most comfortable for them
Further InMomentrsquos highly accurate translation means Tiffany amp Co may use any and every platform featuremdashon
feedback collected in all languagesmdashto sift through comments and research topics most important to the brand
With more than 15 million pieces of customer feedback collected each year comments are used to recognize
exceptional locations and employees identify areas of opportunity and understand the specific drivers of customer
satisfaction The ability to analyze feedback across 18 different languages helps Tiffany amp Co understand nuances
in the experiences and expectations of its customers across each region market and location
013
13
GET TING GLOBAL WITH TRANSL ATION
Human Machine and MoreAs we mentioned in the last section the translation method you choose
can affect the goals you set for the scalability quality actionability and
speed of your text analytics solution
014
Human TranslationHuman translation is the method of having each comment translated
individually as it is submitted by the customer This type of manual
translation requires a long expensive process
Instead of real-time results there is usually a wait of 24 to 48 hours (or
more) to get the translated results And because human translation is
limited to languages that a specific translator can work with outsourcing
more complex languages to other translators will also add additional
costsmdashand more time
LETrsquoS GO OVER THE DIFFERENT OPTIONS
AND THEIR PROS AND CONS
015
Native Language LibrariesA native language library is a text analytics model created
entirely in a language of choice so the system analyzes and
tags comments in that native language instead of translating
them to English tagging them and exporting them back to the
original language Creating a native language library involves a
native speaker building out the initial text analytics model and
then constantly refining it to ensure that it remains up to date
Creating language libraries is a slow and expensive process
usually costing hundreds of thousands of dollars and taking
months or years to build Large amounts of data must be
collected collated and categorized in the language of choice
before the libraries can be built In addition each of these
libraries must be built and tuned to individual industries to
ensure their accuracy The main problem with these native
language libraries therefore is the lack of scalability
016
Machine Translated Human LedMachine translation is exactly what it sounds like It uses
machinesmdashtrained and monitored by humansmdashto translate
text When translating machine translation services look for
patterns and trends in millions of documents to provide the
best translation
Additionally many machine translation services are now using
a neural machine translation engine This means that rather
than translating individual words it uses a neural network
capable of deep learning to look at whole sentences and
understand the greater context These translation services are
highly accurate actionable cost efficient and scalable which
makes them the best option for an international VoC program
017
17
A ldquoMODELrdquo TEXT ANALY TICS SOLUTION
Tailoring Intelligence for Your BusinessThe next element of a text analytics solution to explore is the models you can employ to label
and organize your unstructured data These models have truly come a long way when it comes to their ability to deliver intelligence that is specifically tailored to your business so you can identify
areas that require action and deliver direct value to your bottom line
Traditionally brands were faced with the choice between two specific types of text analytics models industry models and custom models The definition of these two historical approaches
is self explanatory but letrsquos take a deeper dive
018
Industry ModelsAn industry model is developed by experts to reflect the
characteristics of a specific industry These models are
often designed by a vendor to be employed for every client
that fits that particular industry and therefore it often
misses brand-specific insights Additionally because the
same model is used for multiple brands the model canrsquot
be adapted for one brand without being adapted for all
However these models are a much more time-and-cost
effective solution when compared to custom models
Custom ModelsA custom model is built specifically for a certain company
gaining individualized context on the moments that set
a brand apart from its competitors While they are highly
accurate these models are more expensive to create and
maintain so they arenrsquot updated regularly Additionally they
require a high number of comments to test
019
Custom Layered ModelsThere are pros and cons to both of these historical
approaches but there is a new more effective
approach on the scene that gives experience
programs the best of both worlds custom layered
models Custom layered models take the accuracy of
custom models and the efficiency of industry models
to provide tailored accurate and effective results at a
fraction of the cost
How InMomentrsquos Custom Layered Models leverage
over 50 industry models to craft a personalized
tailored solution for each brand And when your text
analytics are built to suit your specific needs the
intelligence you gain from them is multiplied Take a
look at the difference
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retailecommerce
pharmacycustom tags
3xMore Data Points
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retail
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
SINGLE INDUSTRY MODEL CUSTOM L AYERED MODEL
019
020
Comparing and Contrasting ModelsWant a straightforward look at how historical industry and custom models compare to custom layed models
Wersquove laid out the features and benefits of each in this chart
INDUSTRY MODELS CUSTOM MODELS CUSTOM LAYERED MODELS
Covers All Industries
Reflects Unique Business
Delivers Insights for a Combination of Offerings Channels amp Services
Can Be Adapted for an Individual Business without Changing the Model for All Clients Using That Model
Highly Accurate
Cost Effective
Time Efficient
Updated Regularly
Can Be Tested with Limited Data
020
021
21
FACTORING IN SENTIMENT
Getting to the WhyWhen humans look to tell effective storiesmdashwhether in film or in conversationmdashsentiment is
absolutely essential Itrsquos no different when your customer or employee is trying to tell you a
story about their experience via unstructured feedback Itrsquos your text analytics solutionrsquos job to
detect the emotion in their story identify it and help you understand what is causing that specific
sentiment so you can improve experiences But the specific method you use to help your solution
detect sentiment can affect the results you get and therefore your ability to act and transform
022
Rules-Based SentimentTraditionally the way text analytics identified senti-
ment was through a rules-based approach As the name
suggests this method applied rigid rules to understand
unstructured feedback Additionally changing these
rules or updating them required manual intervention
And even with those manual updates accuracy never
really improves
Adaptive Sentiment EngineThe idea that inspired the adaptive sentiment engine
approach is that rigid rules do not lend themselves to best
understand human emotions Instead adaptive sentiment
is AI driven meaning that it learns and improves over time
giving you more accurate results This engine continually
recognizes new terms and phrases as well meaning that
it grows and adapts in lockstep with your brand Check out
the difference below
023
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive Positive Negative NegativeSentiment Score 92
Confidence 98Sentiment Score 52
Confidence 97Sentiment Score -49
Confidence 41Sentiment Score -79
Confidence 94
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
RULES -BASED MACHINE LEARNING
NegativeNeutral
024
Now that you know what it takes to make your text analytics
solution truly world-class what next Well itrsquos time to make
sure that your greater experience program is set up to actually
drive improvement for your organization
Text analytics are just one piece of the puzzle when it comes
to using your VoC VoE or other experience initiatives to
power business growth When you look at the bigger picture
it becomes clear that you need a solid strategy for success
At InMoment we leverage the Continuous Improvement
Framework a method that helps us guide our clients through
designing listening understanding transforming and realizing
their goals for their experience initiative
If you want to learn more about this framework
and how text analytics fit into it you can read
more here
Like what yoursquove seen so far
Here are a few other pieces you may like
How You Listen Matters Modernizing Your Methods amp Approach to Customer Feedback
How to Prove the Business Value of Your CX Program
Building a Business Case for Your CFO (or Anyone Else)
Measuring Doesnrsquot Make You Taller
Therersquos more where that came from on
wwwinmomentcomresources
DRIVING IMPROVEMENT
024
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C
09
09
TOP OBJECTIVES FOR A MODERN TEXT ANALY TICS SOLUTION
Scalability Quality Actionability and SpeedSo if 100 accuracy is neither achievable nor the end-all-be-all of text analytics solutions what
should experience programs strive for Here are four key areas to consider when measuring the
effectiveness of a text analytics solution
010
Scalability
Any and every growing business (especially those with
a global reach) needs a solution that supports all of the
countries and languages your customers work and buy
inmdashat an acceptable level of quality and at a price that
makes sense No one has unlimited resources so the
first criteria must be that the text analytics solution you
choose is scalable
Quality
The goal of text analytics is not to allow you to read
every single comment that your customers submit
For most businesses this is simply not feasible unless
you have a team of analysts dedicated to it full time
Your text analytics solution must be able to understand
and surface important trends and patterns based on
individual comments and the sentiments behind them
011
Actionability
Another key objective of text analytics is to be able to
take action on the customer issues that most affect your
business This means you need a layer of sophisticated
analytics that can add tags and themes on a granular level
uncover sentiment assign categories identify intent spot
legal issues and possible customer churn and more
Speed
Business today moves quickly so you need a text analytics
solution that can keep pacemdashwhich often means real-time
analysis and reporting This is specifically relevant when
considering translations for global companies The need
for speed means that brands donrsquot have time to wait for a
perfect translation of each and every comment because
perfect translations are only possible if a team manually
translates every word which takes a significant and
unrealistic amount of time amp money
Scalability
Any and every growing business (especially those with
a global reach) needs a solution that supports all of the
countries and languages your customers work and buy
inmdashat an acceptable level of quality and at a price that
makes sense No one has unlimited resources so the
first criteria must be that the text analytics solution you
choose is scalable
012
Tiffany amp Co Leverages Global Text Analytics for Market-Specific and Company-Wide Understanding of Experiences
For a global company the ability to collect and analyze customer feedback across multiple regions and languagesmdash
within a single platformmdashis essential Tiffany amp Co one of the worldrsquos most recognizable and respected luxury
brands leverages InMoment in 18 languages across 300 company-owned stores multiple authorized retailers
its e-commerce platform and in its customer care center for a real-time understanding of market-specific and
company-wide feedback trends
After translation to English comments are analyzed using approximately 150 custom tags and customer sentiment
is displayed in a multitude of reports and dashboards in both English and the original comment language This
way both local and corporate program owners can view results in the language that is most comfortable for them
Further InMomentrsquos highly accurate translation means Tiffany amp Co may use any and every platform featuremdashon
feedback collected in all languagesmdashto sift through comments and research topics most important to the brand
With more than 15 million pieces of customer feedback collected each year comments are used to recognize
exceptional locations and employees identify areas of opportunity and understand the specific drivers of customer
satisfaction The ability to analyze feedback across 18 different languages helps Tiffany amp Co understand nuances
in the experiences and expectations of its customers across each region market and location
013
13
GET TING GLOBAL WITH TRANSL ATION
Human Machine and MoreAs we mentioned in the last section the translation method you choose
can affect the goals you set for the scalability quality actionability and
speed of your text analytics solution
014
Human TranslationHuman translation is the method of having each comment translated
individually as it is submitted by the customer This type of manual
translation requires a long expensive process
Instead of real-time results there is usually a wait of 24 to 48 hours (or
more) to get the translated results And because human translation is
limited to languages that a specific translator can work with outsourcing
more complex languages to other translators will also add additional
costsmdashand more time
LETrsquoS GO OVER THE DIFFERENT OPTIONS
AND THEIR PROS AND CONS
015
Native Language LibrariesA native language library is a text analytics model created
entirely in a language of choice so the system analyzes and
tags comments in that native language instead of translating
them to English tagging them and exporting them back to the
original language Creating a native language library involves a
native speaker building out the initial text analytics model and
then constantly refining it to ensure that it remains up to date
Creating language libraries is a slow and expensive process
usually costing hundreds of thousands of dollars and taking
months or years to build Large amounts of data must be
collected collated and categorized in the language of choice
before the libraries can be built In addition each of these
libraries must be built and tuned to individual industries to
ensure their accuracy The main problem with these native
language libraries therefore is the lack of scalability
016
Machine Translated Human LedMachine translation is exactly what it sounds like It uses
machinesmdashtrained and monitored by humansmdashto translate
text When translating machine translation services look for
patterns and trends in millions of documents to provide the
best translation
Additionally many machine translation services are now using
a neural machine translation engine This means that rather
than translating individual words it uses a neural network
capable of deep learning to look at whole sentences and
understand the greater context These translation services are
highly accurate actionable cost efficient and scalable which
makes them the best option for an international VoC program
017
17
A ldquoMODELrdquo TEXT ANALY TICS SOLUTION
Tailoring Intelligence for Your BusinessThe next element of a text analytics solution to explore is the models you can employ to label
and organize your unstructured data These models have truly come a long way when it comes to their ability to deliver intelligence that is specifically tailored to your business so you can identify
areas that require action and deliver direct value to your bottom line
Traditionally brands were faced with the choice between two specific types of text analytics models industry models and custom models The definition of these two historical approaches
is self explanatory but letrsquos take a deeper dive
018
Industry ModelsAn industry model is developed by experts to reflect the
characteristics of a specific industry These models are
often designed by a vendor to be employed for every client
that fits that particular industry and therefore it often
misses brand-specific insights Additionally because the
same model is used for multiple brands the model canrsquot
be adapted for one brand without being adapted for all
However these models are a much more time-and-cost
effective solution when compared to custom models
Custom ModelsA custom model is built specifically for a certain company
gaining individualized context on the moments that set
a brand apart from its competitors While they are highly
accurate these models are more expensive to create and
maintain so they arenrsquot updated regularly Additionally they
require a high number of comments to test
019
Custom Layered ModelsThere are pros and cons to both of these historical
approaches but there is a new more effective
approach on the scene that gives experience
programs the best of both worlds custom layered
models Custom layered models take the accuracy of
custom models and the efficiency of industry models
to provide tailored accurate and effective results at a
fraction of the cost
How InMomentrsquos Custom Layered Models leverage
over 50 industry models to craft a personalized
tailored solution for each brand And when your text
analytics are built to suit your specific needs the
intelligence you gain from them is multiplied Take a
look at the difference
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retailecommerce
pharmacycustom tags
3xMore Data Points
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retail
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
SINGLE INDUSTRY MODEL CUSTOM L AYERED MODEL
019
020
Comparing and Contrasting ModelsWant a straightforward look at how historical industry and custom models compare to custom layed models
Wersquove laid out the features and benefits of each in this chart
INDUSTRY MODELS CUSTOM MODELS CUSTOM LAYERED MODELS
Covers All Industries
Reflects Unique Business
Delivers Insights for a Combination of Offerings Channels amp Services
Can Be Adapted for an Individual Business without Changing the Model for All Clients Using That Model
Highly Accurate
Cost Effective
Time Efficient
Updated Regularly
Can Be Tested with Limited Data
020
021
21
FACTORING IN SENTIMENT
Getting to the WhyWhen humans look to tell effective storiesmdashwhether in film or in conversationmdashsentiment is
absolutely essential Itrsquos no different when your customer or employee is trying to tell you a
story about their experience via unstructured feedback Itrsquos your text analytics solutionrsquos job to
detect the emotion in their story identify it and help you understand what is causing that specific
sentiment so you can improve experiences But the specific method you use to help your solution
detect sentiment can affect the results you get and therefore your ability to act and transform
022
Rules-Based SentimentTraditionally the way text analytics identified senti-
ment was through a rules-based approach As the name
suggests this method applied rigid rules to understand
unstructured feedback Additionally changing these
rules or updating them required manual intervention
And even with those manual updates accuracy never
really improves
Adaptive Sentiment EngineThe idea that inspired the adaptive sentiment engine
approach is that rigid rules do not lend themselves to best
understand human emotions Instead adaptive sentiment
is AI driven meaning that it learns and improves over time
giving you more accurate results This engine continually
recognizes new terms and phrases as well meaning that
it grows and adapts in lockstep with your brand Check out
the difference below
023
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive Positive Negative NegativeSentiment Score 92
Confidence 98Sentiment Score 52
Confidence 97Sentiment Score -49
Confidence 41Sentiment Score -79
Confidence 94
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
RULES -BASED MACHINE LEARNING
NegativeNeutral
024
Now that you know what it takes to make your text analytics
solution truly world-class what next Well itrsquos time to make
sure that your greater experience program is set up to actually
drive improvement for your organization
Text analytics are just one piece of the puzzle when it comes
to using your VoC VoE or other experience initiatives to
power business growth When you look at the bigger picture
it becomes clear that you need a solid strategy for success
At InMoment we leverage the Continuous Improvement
Framework a method that helps us guide our clients through
designing listening understanding transforming and realizing
their goals for their experience initiative
If you want to learn more about this framework
and how text analytics fit into it you can read
more here
Like what yoursquove seen so far
Here are a few other pieces you may like
How You Listen Matters Modernizing Your Methods amp Approach to Customer Feedback
How to Prove the Business Value of Your CX Program
Building a Business Case for Your CFO (or Anyone Else)
Measuring Doesnrsquot Make You Taller
Therersquos more where that came from on
wwwinmomentcomresources
DRIVING IMPROVEMENT
024
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C
010
Scalability
Any and every growing business (especially those with
a global reach) needs a solution that supports all of the
countries and languages your customers work and buy
inmdashat an acceptable level of quality and at a price that
makes sense No one has unlimited resources so the
first criteria must be that the text analytics solution you
choose is scalable
Quality
The goal of text analytics is not to allow you to read
every single comment that your customers submit
For most businesses this is simply not feasible unless
you have a team of analysts dedicated to it full time
Your text analytics solution must be able to understand
and surface important trends and patterns based on
individual comments and the sentiments behind them
011
Actionability
Another key objective of text analytics is to be able to
take action on the customer issues that most affect your
business This means you need a layer of sophisticated
analytics that can add tags and themes on a granular level
uncover sentiment assign categories identify intent spot
legal issues and possible customer churn and more
Speed
Business today moves quickly so you need a text analytics
solution that can keep pacemdashwhich often means real-time
analysis and reporting This is specifically relevant when
considering translations for global companies The need
for speed means that brands donrsquot have time to wait for a
perfect translation of each and every comment because
perfect translations are only possible if a team manually
translates every word which takes a significant and
unrealistic amount of time amp money
Scalability
Any and every growing business (especially those with
a global reach) needs a solution that supports all of the
countries and languages your customers work and buy
inmdashat an acceptable level of quality and at a price that
makes sense No one has unlimited resources so the
first criteria must be that the text analytics solution you
choose is scalable
012
Tiffany amp Co Leverages Global Text Analytics for Market-Specific and Company-Wide Understanding of Experiences
For a global company the ability to collect and analyze customer feedback across multiple regions and languagesmdash
within a single platformmdashis essential Tiffany amp Co one of the worldrsquos most recognizable and respected luxury
brands leverages InMoment in 18 languages across 300 company-owned stores multiple authorized retailers
its e-commerce platform and in its customer care center for a real-time understanding of market-specific and
company-wide feedback trends
After translation to English comments are analyzed using approximately 150 custom tags and customer sentiment
is displayed in a multitude of reports and dashboards in both English and the original comment language This
way both local and corporate program owners can view results in the language that is most comfortable for them
Further InMomentrsquos highly accurate translation means Tiffany amp Co may use any and every platform featuremdashon
feedback collected in all languagesmdashto sift through comments and research topics most important to the brand
With more than 15 million pieces of customer feedback collected each year comments are used to recognize
exceptional locations and employees identify areas of opportunity and understand the specific drivers of customer
satisfaction The ability to analyze feedback across 18 different languages helps Tiffany amp Co understand nuances
in the experiences and expectations of its customers across each region market and location
013
13
GET TING GLOBAL WITH TRANSL ATION
Human Machine and MoreAs we mentioned in the last section the translation method you choose
can affect the goals you set for the scalability quality actionability and
speed of your text analytics solution
014
Human TranslationHuman translation is the method of having each comment translated
individually as it is submitted by the customer This type of manual
translation requires a long expensive process
Instead of real-time results there is usually a wait of 24 to 48 hours (or
more) to get the translated results And because human translation is
limited to languages that a specific translator can work with outsourcing
more complex languages to other translators will also add additional
costsmdashand more time
LETrsquoS GO OVER THE DIFFERENT OPTIONS
AND THEIR PROS AND CONS
015
Native Language LibrariesA native language library is a text analytics model created
entirely in a language of choice so the system analyzes and
tags comments in that native language instead of translating
them to English tagging them and exporting them back to the
original language Creating a native language library involves a
native speaker building out the initial text analytics model and
then constantly refining it to ensure that it remains up to date
Creating language libraries is a slow and expensive process
usually costing hundreds of thousands of dollars and taking
months or years to build Large amounts of data must be
collected collated and categorized in the language of choice
before the libraries can be built In addition each of these
libraries must be built and tuned to individual industries to
ensure their accuracy The main problem with these native
language libraries therefore is the lack of scalability
016
Machine Translated Human LedMachine translation is exactly what it sounds like It uses
machinesmdashtrained and monitored by humansmdashto translate
text When translating machine translation services look for
patterns and trends in millions of documents to provide the
best translation
Additionally many machine translation services are now using
a neural machine translation engine This means that rather
than translating individual words it uses a neural network
capable of deep learning to look at whole sentences and
understand the greater context These translation services are
highly accurate actionable cost efficient and scalable which
makes them the best option for an international VoC program
017
17
A ldquoMODELrdquo TEXT ANALY TICS SOLUTION
Tailoring Intelligence for Your BusinessThe next element of a text analytics solution to explore is the models you can employ to label
and organize your unstructured data These models have truly come a long way when it comes to their ability to deliver intelligence that is specifically tailored to your business so you can identify
areas that require action and deliver direct value to your bottom line
Traditionally brands were faced with the choice between two specific types of text analytics models industry models and custom models The definition of these two historical approaches
is self explanatory but letrsquos take a deeper dive
018
Industry ModelsAn industry model is developed by experts to reflect the
characteristics of a specific industry These models are
often designed by a vendor to be employed for every client
that fits that particular industry and therefore it often
misses brand-specific insights Additionally because the
same model is used for multiple brands the model canrsquot
be adapted for one brand without being adapted for all
However these models are a much more time-and-cost
effective solution when compared to custom models
Custom ModelsA custom model is built specifically for a certain company
gaining individualized context on the moments that set
a brand apart from its competitors While they are highly
accurate these models are more expensive to create and
maintain so they arenrsquot updated regularly Additionally they
require a high number of comments to test
019
Custom Layered ModelsThere are pros and cons to both of these historical
approaches but there is a new more effective
approach on the scene that gives experience
programs the best of both worlds custom layered
models Custom layered models take the accuracy of
custom models and the efficiency of industry models
to provide tailored accurate and effective results at a
fraction of the cost
How InMomentrsquos Custom Layered Models leverage
over 50 industry models to craft a personalized
tailored solution for each brand And when your text
analytics are built to suit your specific needs the
intelligence you gain from them is multiplied Take a
look at the difference
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retailecommerce
pharmacycustom tags
3xMore Data Points
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retail
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
SINGLE INDUSTRY MODEL CUSTOM L AYERED MODEL
019
020
Comparing and Contrasting ModelsWant a straightforward look at how historical industry and custom models compare to custom layed models
Wersquove laid out the features and benefits of each in this chart
INDUSTRY MODELS CUSTOM MODELS CUSTOM LAYERED MODELS
Covers All Industries
Reflects Unique Business
Delivers Insights for a Combination of Offerings Channels amp Services
Can Be Adapted for an Individual Business without Changing the Model for All Clients Using That Model
Highly Accurate
Cost Effective
Time Efficient
Updated Regularly
Can Be Tested with Limited Data
020
021
21
FACTORING IN SENTIMENT
Getting to the WhyWhen humans look to tell effective storiesmdashwhether in film or in conversationmdashsentiment is
absolutely essential Itrsquos no different when your customer or employee is trying to tell you a
story about their experience via unstructured feedback Itrsquos your text analytics solutionrsquos job to
detect the emotion in their story identify it and help you understand what is causing that specific
sentiment so you can improve experiences But the specific method you use to help your solution
detect sentiment can affect the results you get and therefore your ability to act and transform
022
Rules-Based SentimentTraditionally the way text analytics identified senti-
ment was through a rules-based approach As the name
suggests this method applied rigid rules to understand
unstructured feedback Additionally changing these
rules or updating them required manual intervention
And even with those manual updates accuracy never
really improves
Adaptive Sentiment EngineThe idea that inspired the adaptive sentiment engine
approach is that rigid rules do not lend themselves to best
understand human emotions Instead adaptive sentiment
is AI driven meaning that it learns and improves over time
giving you more accurate results This engine continually
recognizes new terms and phrases as well meaning that
it grows and adapts in lockstep with your brand Check out
the difference below
023
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive Positive Negative NegativeSentiment Score 92
Confidence 98Sentiment Score 52
Confidence 97Sentiment Score -49
Confidence 41Sentiment Score -79
Confidence 94
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
RULES -BASED MACHINE LEARNING
NegativeNeutral
024
Now that you know what it takes to make your text analytics
solution truly world-class what next Well itrsquos time to make
sure that your greater experience program is set up to actually
drive improvement for your organization
Text analytics are just one piece of the puzzle when it comes
to using your VoC VoE or other experience initiatives to
power business growth When you look at the bigger picture
it becomes clear that you need a solid strategy for success
At InMoment we leverage the Continuous Improvement
Framework a method that helps us guide our clients through
designing listening understanding transforming and realizing
their goals for their experience initiative
If you want to learn more about this framework
and how text analytics fit into it you can read
more here
Like what yoursquove seen so far
Here are a few other pieces you may like
How You Listen Matters Modernizing Your Methods amp Approach to Customer Feedback
How to Prove the Business Value of Your CX Program
Building a Business Case for Your CFO (or Anyone Else)
Measuring Doesnrsquot Make You Taller
Therersquos more where that came from on
wwwinmomentcomresources
DRIVING IMPROVEMENT
024
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C
011
Actionability
Another key objective of text analytics is to be able to
take action on the customer issues that most affect your
business This means you need a layer of sophisticated
analytics that can add tags and themes on a granular level
uncover sentiment assign categories identify intent spot
legal issues and possible customer churn and more
Speed
Business today moves quickly so you need a text analytics
solution that can keep pacemdashwhich often means real-time
analysis and reporting This is specifically relevant when
considering translations for global companies The need
for speed means that brands donrsquot have time to wait for a
perfect translation of each and every comment because
perfect translations are only possible if a team manually
translates every word which takes a significant and
unrealistic amount of time amp money
Scalability
Any and every growing business (especially those with
a global reach) needs a solution that supports all of the
countries and languages your customers work and buy
inmdashat an acceptable level of quality and at a price that
makes sense No one has unlimited resources so the
first criteria must be that the text analytics solution you
choose is scalable
012
Tiffany amp Co Leverages Global Text Analytics for Market-Specific and Company-Wide Understanding of Experiences
For a global company the ability to collect and analyze customer feedback across multiple regions and languagesmdash
within a single platformmdashis essential Tiffany amp Co one of the worldrsquos most recognizable and respected luxury
brands leverages InMoment in 18 languages across 300 company-owned stores multiple authorized retailers
its e-commerce platform and in its customer care center for a real-time understanding of market-specific and
company-wide feedback trends
After translation to English comments are analyzed using approximately 150 custom tags and customer sentiment
is displayed in a multitude of reports and dashboards in both English and the original comment language This
way both local and corporate program owners can view results in the language that is most comfortable for them
Further InMomentrsquos highly accurate translation means Tiffany amp Co may use any and every platform featuremdashon
feedback collected in all languagesmdashto sift through comments and research topics most important to the brand
With more than 15 million pieces of customer feedback collected each year comments are used to recognize
exceptional locations and employees identify areas of opportunity and understand the specific drivers of customer
satisfaction The ability to analyze feedback across 18 different languages helps Tiffany amp Co understand nuances
in the experiences and expectations of its customers across each region market and location
013
13
GET TING GLOBAL WITH TRANSL ATION
Human Machine and MoreAs we mentioned in the last section the translation method you choose
can affect the goals you set for the scalability quality actionability and
speed of your text analytics solution
014
Human TranslationHuman translation is the method of having each comment translated
individually as it is submitted by the customer This type of manual
translation requires a long expensive process
Instead of real-time results there is usually a wait of 24 to 48 hours (or
more) to get the translated results And because human translation is
limited to languages that a specific translator can work with outsourcing
more complex languages to other translators will also add additional
costsmdashand more time
LETrsquoS GO OVER THE DIFFERENT OPTIONS
AND THEIR PROS AND CONS
015
Native Language LibrariesA native language library is a text analytics model created
entirely in a language of choice so the system analyzes and
tags comments in that native language instead of translating
them to English tagging them and exporting them back to the
original language Creating a native language library involves a
native speaker building out the initial text analytics model and
then constantly refining it to ensure that it remains up to date
Creating language libraries is a slow and expensive process
usually costing hundreds of thousands of dollars and taking
months or years to build Large amounts of data must be
collected collated and categorized in the language of choice
before the libraries can be built In addition each of these
libraries must be built and tuned to individual industries to
ensure their accuracy The main problem with these native
language libraries therefore is the lack of scalability
016
Machine Translated Human LedMachine translation is exactly what it sounds like It uses
machinesmdashtrained and monitored by humansmdashto translate
text When translating machine translation services look for
patterns and trends in millions of documents to provide the
best translation
Additionally many machine translation services are now using
a neural machine translation engine This means that rather
than translating individual words it uses a neural network
capable of deep learning to look at whole sentences and
understand the greater context These translation services are
highly accurate actionable cost efficient and scalable which
makes them the best option for an international VoC program
017
17
A ldquoMODELrdquo TEXT ANALY TICS SOLUTION
Tailoring Intelligence for Your BusinessThe next element of a text analytics solution to explore is the models you can employ to label
and organize your unstructured data These models have truly come a long way when it comes to their ability to deliver intelligence that is specifically tailored to your business so you can identify
areas that require action and deliver direct value to your bottom line
Traditionally brands were faced with the choice between two specific types of text analytics models industry models and custom models The definition of these two historical approaches
is self explanatory but letrsquos take a deeper dive
018
Industry ModelsAn industry model is developed by experts to reflect the
characteristics of a specific industry These models are
often designed by a vendor to be employed for every client
that fits that particular industry and therefore it often
misses brand-specific insights Additionally because the
same model is used for multiple brands the model canrsquot
be adapted for one brand without being adapted for all
However these models are a much more time-and-cost
effective solution when compared to custom models
Custom ModelsA custom model is built specifically for a certain company
gaining individualized context on the moments that set
a brand apart from its competitors While they are highly
accurate these models are more expensive to create and
maintain so they arenrsquot updated regularly Additionally they
require a high number of comments to test
019
Custom Layered ModelsThere are pros and cons to both of these historical
approaches but there is a new more effective
approach on the scene that gives experience
programs the best of both worlds custom layered
models Custom layered models take the accuracy of
custom models and the efficiency of industry models
to provide tailored accurate and effective results at a
fraction of the cost
How InMomentrsquos Custom Layered Models leverage
over 50 industry models to craft a personalized
tailored solution for each brand And when your text
analytics are built to suit your specific needs the
intelligence you gain from them is multiplied Take a
look at the difference
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retailecommerce
pharmacycustom tags
3xMore Data Points
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retail
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
SINGLE INDUSTRY MODEL CUSTOM L AYERED MODEL
019
020
Comparing and Contrasting ModelsWant a straightforward look at how historical industry and custom models compare to custom layed models
Wersquove laid out the features and benefits of each in this chart
INDUSTRY MODELS CUSTOM MODELS CUSTOM LAYERED MODELS
Covers All Industries
Reflects Unique Business
Delivers Insights for a Combination of Offerings Channels amp Services
Can Be Adapted for an Individual Business without Changing the Model for All Clients Using That Model
Highly Accurate
Cost Effective
Time Efficient
Updated Regularly
Can Be Tested with Limited Data
020
021
21
FACTORING IN SENTIMENT
Getting to the WhyWhen humans look to tell effective storiesmdashwhether in film or in conversationmdashsentiment is
absolutely essential Itrsquos no different when your customer or employee is trying to tell you a
story about their experience via unstructured feedback Itrsquos your text analytics solutionrsquos job to
detect the emotion in their story identify it and help you understand what is causing that specific
sentiment so you can improve experiences But the specific method you use to help your solution
detect sentiment can affect the results you get and therefore your ability to act and transform
022
Rules-Based SentimentTraditionally the way text analytics identified senti-
ment was through a rules-based approach As the name
suggests this method applied rigid rules to understand
unstructured feedback Additionally changing these
rules or updating them required manual intervention
And even with those manual updates accuracy never
really improves
Adaptive Sentiment EngineThe idea that inspired the adaptive sentiment engine
approach is that rigid rules do not lend themselves to best
understand human emotions Instead adaptive sentiment
is AI driven meaning that it learns and improves over time
giving you more accurate results This engine continually
recognizes new terms and phrases as well meaning that
it grows and adapts in lockstep with your brand Check out
the difference below
023
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive Positive Negative NegativeSentiment Score 92
Confidence 98Sentiment Score 52
Confidence 97Sentiment Score -49
Confidence 41Sentiment Score -79
Confidence 94
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
RULES -BASED MACHINE LEARNING
NegativeNeutral
024
Now that you know what it takes to make your text analytics
solution truly world-class what next Well itrsquos time to make
sure that your greater experience program is set up to actually
drive improvement for your organization
Text analytics are just one piece of the puzzle when it comes
to using your VoC VoE or other experience initiatives to
power business growth When you look at the bigger picture
it becomes clear that you need a solid strategy for success
At InMoment we leverage the Continuous Improvement
Framework a method that helps us guide our clients through
designing listening understanding transforming and realizing
their goals for their experience initiative
If you want to learn more about this framework
and how text analytics fit into it you can read
more here
Like what yoursquove seen so far
Here are a few other pieces you may like
How You Listen Matters Modernizing Your Methods amp Approach to Customer Feedback
How to Prove the Business Value of Your CX Program
Building a Business Case for Your CFO (or Anyone Else)
Measuring Doesnrsquot Make You Taller
Therersquos more where that came from on
wwwinmomentcomresources
DRIVING IMPROVEMENT
024
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C
012
Tiffany amp Co Leverages Global Text Analytics for Market-Specific and Company-Wide Understanding of Experiences
For a global company the ability to collect and analyze customer feedback across multiple regions and languagesmdash
within a single platformmdashis essential Tiffany amp Co one of the worldrsquos most recognizable and respected luxury
brands leverages InMoment in 18 languages across 300 company-owned stores multiple authorized retailers
its e-commerce platform and in its customer care center for a real-time understanding of market-specific and
company-wide feedback trends
After translation to English comments are analyzed using approximately 150 custom tags and customer sentiment
is displayed in a multitude of reports and dashboards in both English and the original comment language This
way both local and corporate program owners can view results in the language that is most comfortable for them
Further InMomentrsquos highly accurate translation means Tiffany amp Co may use any and every platform featuremdashon
feedback collected in all languagesmdashto sift through comments and research topics most important to the brand
With more than 15 million pieces of customer feedback collected each year comments are used to recognize
exceptional locations and employees identify areas of opportunity and understand the specific drivers of customer
satisfaction The ability to analyze feedback across 18 different languages helps Tiffany amp Co understand nuances
in the experiences and expectations of its customers across each region market and location
013
13
GET TING GLOBAL WITH TRANSL ATION
Human Machine and MoreAs we mentioned in the last section the translation method you choose
can affect the goals you set for the scalability quality actionability and
speed of your text analytics solution
014
Human TranslationHuman translation is the method of having each comment translated
individually as it is submitted by the customer This type of manual
translation requires a long expensive process
Instead of real-time results there is usually a wait of 24 to 48 hours (or
more) to get the translated results And because human translation is
limited to languages that a specific translator can work with outsourcing
more complex languages to other translators will also add additional
costsmdashand more time
LETrsquoS GO OVER THE DIFFERENT OPTIONS
AND THEIR PROS AND CONS
015
Native Language LibrariesA native language library is a text analytics model created
entirely in a language of choice so the system analyzes and
tags comments in that native language instead of translating
them to English tagging them and exporting them back to the
original language Creating a native language library involves a
native speaker building out the initial text analytics model and
then constantly refining it to ensure that it remains up to date
Creating language libraries is a slow and expensive process
usually costing hundreds of thousands of dollars and taking
months or years to build Large amounts of data must be
collected collated and categorized in the language of choice
before the libraries can be built In addition each of these
libraries must be built and tuned to individual industries to
ensure their accuracy The main problem with these native
language libraries therefore is the lack of scalability
016
Machine Translated Human LedMachine translation is exactly what it sounds like It uses
machinesmdashtrained and monitored by humansmdashto translate
text When translating machine translation services look for
patterns and trends in millions of documents to provide the
best translation
Additionally many machine translation services are now using
a neural machine translation engine This means that rather
than translating individual words it uses a neural network
capable of deep learning to look at whole sentences and
understand the greater context These translation services are
highly accurate actionable cost efficient and scalable which
makes them the best option for an international VoC program
017
17
A ldquoMODELrdquo TEXT ANALY TICS SOLUTION
Tailoring Intelligence for Your BusinessThe next element of a text analytics solution to explore is the models you can employ to label
and organize your unstructured data These models have truly come a long way when it comes to their ability to deliver intelligence that is specifically tailored to your business so you can identify
areas that require action and deliver direct value to your bottom line
Traditionally brands were faced with the choice between two specific types of text analytics models industry models and custom models The definition of these two historical approaches
is self explanatory but letrsquos take a deeper dive
018
Industry ModelsAn industry model is developed by experts to reflect the
characteristics of a specific industry These models are
often designed by a vendor to be employed for every client
that fits that particular industry and therefore it often
misses brand-specific insights Additionally because the
same model is used for multiple brands the model canrsquot
be adapted for one brand without being adapted for all
However these models are a much more time-and-cost
effective solution when compared to custom models
Custom ModelsA custom model is built specifically for a certain company
gaining individualized context on the moments that set
a brand apart from its competitors While they are highly
accurate these models are more expensive to create and
maintain so they arenrsquot updated regularly Additionally they
require a high number of comments to test
019
Custom Layered ModelsThere are pros and cons to both of these historical
approaches but there is a new more effective
approach on the scene that gives experience
programs the best of both worlds custom layered
models Custom layered models take the accuracy of
custom models and the efficiency of industry models
to provide tailored accurate and effective results at a
fraction of the cost
How InMomentrsquos Custom Layered Models leverage
over 50 industry models to craft a personalized
tailored solution for each brand And when your text
analytics are built to suit your specific needs the
intelligence you gain from them is multiplied Take a
look at the difference
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retailecommerce
pharmacycustom tags
3xMore Data Points
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retail
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
SINGLE INDUSTRY MODEL CUSTOM L AYERED MODEL
019
020
Comparing and Contrasting ModelsWant a straightforward look at how historical industry and custom models compare to custom layed models
Wersquove laid out the features and benefits of each in this chart
INDUSTRY MODELS CUSTOM MODELS CUSTOM LAYERED MODELS
Covers All Industries
Reflects Unique Business
Delivers Insights for a Combination of Offerings Channels amp Services
Can Be Adapted for an Individual Business without Changing the Model for All Clients Using That Model
Highly Accurate
Cost Effective
Time Efficient
Updated Regularly
Can Be Tested with Limited Data
020
021
21
FACTORING IN SENTIMENT
Getting to the WhyWhen humans look to tell effective storiesmdashwhether in film or in conversationmdashsentiment is
absolutely essential Itrsquos no different when your customer or employee is trying to tell you a
story about their experience via unstructured feedback Itrsquos your text analytics solutionrsquos job to
detect the emotion in their story identify it and help you understand what is causing that specific
sentiment so you can improve experiences But the specific method you use to help your solution
detect sentiment can affect the results you get and therefore your ability to act and transform
022
Rules-Based SentimentTraditionally the way text analytics identified senti-
ment was through a rules-based approach As the name
suggests this method applied rigid rules to understand
unstructured feedback Additionally changing these
rules or updating them required manual intervention
And even with those manual updates accuracy never
really improves
Adaptive Sentiment EngineThe idea that inspired the adaptive sentiment engine
approach is that rigid rules do not lend themselves to best
understand human emotions Instead adaptive sentiment
is AI driven meaning that it learns and improves over time
giving you more accurate results This engine continually
recognizes new terms and phrases as well meaning that
it grows and adapts in lockstep with your brand Check out
the difference below
023
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive Positive Negative NegativeSentiment Score 92
Confidence 98Sentiment Score 52
Confidence 97Sentiment Score -49
Confidence 41Sentiment Score -79
Confidence 94
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
RULES -BASED MACHINE LEARNING
NegativeNeutral
024
Now that you know what it takes to make your text analytics
solution truly world-class what next Well itrsquos time to make
sure that your greater experience program is set up to actually
drive improvement for your organization
Text analytics are just one piece of the puzzle when it comes
to using your VoC VoE or other experience initiatives to
power business growth When you look at the bigger picture
it becomes clear that you need a solid strategy for success
At InMoment we leverage the Continuous Improvement
Framework a method that helps us guide our clients through
designing listening understanding transforming and realizing
their goals for their experience initiative
If you want to learn more about this framework
and how text analytics fit into it you can read
more here
Like what yoursquove seen so far
Here are a few other pieces you may like
How You Listen Matters Modernizing Your Methods amp Approach to Customer Feedback
How to Prove the Business Value of Your CX Program
Building a Business Case for Your CFO (or Anyone Else)
Measuring Doesnrsquot Make You Taller
Therersquos more where that came from on
wwwinmomentcomresources
DRIVING IMPROVEMENT
024
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C
013
13
GET TING GLOBAL WITH TRANSL ATION
Human Machine and MoreAs we mentioned in the last section the translation method you choose
can affect the goals you set for the scalability quality actionability and
speed of your text analytics solution
014
Human TranslationHuman translation is the method of having each comment translated
individually as it is submitted by the customer This type of manual
translation requires a long expensive process
Instead of real-time results there is usually a wait of 24 to 48 hours (or
more) to get the translated results And because human translation is
limited to languages that a specific translator can work with outsourcing
more complex languages to other translators will also add additional
costsmdashand more time
LETrsquoS GO OVER THE DIFFERENT OPTIONS
AND THEIR PROS AND CONS
015
Native Language LibrariesA native language library is a text analytics model created
entirely in a language of choice so the system analyzes and
tags comments in that native language instead of translating
them to English tagging them and exporting them back to the
original language Creating a native language library involves a
native speaker building out the initial text analytics model and
then constantly refining it to ensure that it remains up to date
Creating language libraries is a slow and expensive process
usually costing hundreds of thousands of dollars and taking
months or years to build Large amounts of data must be
collected collated and categorized in the language of choice
before the libraries can be built In addition each of these
libraries must be built and tuned to individual industries to
ensure their accuracy The main problem with these native
language libraries therefore is the lack of scalability
016
Machine Translated Human LedMachine translation is exactly what it sounds like It uses
machinesmdashtrained and monitored by humansmdashto translate
text When translating machine translation services look for
patterns and trends in millions of documents to provide the
best translation
Additionally many machine translation services are now using
a neural machine translation engine This means that rather
than translating individual words it uses a neural network
capable of deep learning to look at whole sentences and
understand the greater context These translation services are
highly accurate actionable cost efficient and scalable which
makes them the best option for an international VoC program
017
17
A ldquoMODELrdquo TEXT ANALY TICS SOLUTION
Tailoring Intelligence for Your BusinessThe next element of a text analytics solution to explore is the models you can employ to label
and organize your unstructured data These models have truly come a long way when it comes to their ability to deliver intelligence that is specifically tailored to your business so you can identify
areas that require action and deliver direct value to your bottom line
Traditionally brands were faced with the choice between two specific types of text analytics models industry models and custom models The definition of these two historical approaches
is self explanatory but letrsquos take a deeper dive
018
Industry ModelsAn industry model is developed by experts to reflect the
characteristics of a specific industry These models are
often designed by a vendor to be employed for every client
that fits that particular industry and therefore it often
misses brand-specific insights Additionally because the
same model is used for multiple brands the model canrsquot
be adapted for one brand without being adapted for all
However these models are a much more time-and-cost
effective solution when compared to custom models
Custom ModelsA custom model is built specifically for a certain company
gaining individualized context on the moments that set
a brand apart from its competitors While they are highly
accurate these models are more expensive to create and
maintain so they arenrsquot updated regularly Additionally they
require a high number of comments to test
019
Custom Layered ModelsThere are pros and cons to both of these historical
approaches but there is a new more effective
approach on the scene that gives experience
programs the best of both worlds custom layered
models Custom layered models take the accuracy of
custom models and the efficiency of industry models
to provide tailored accurate and effective results at a
fraction of the cost
How InMomentrsquos Custom Layered Models leverage
over 50 industry models to craft a personalized
tailored solution for each brand And when your text
analytics are built to suit your specific needs the
intelligence you gain from them is multiplied Take a
look at the difference
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retailecommerce
pharmacycustom tags
3xMore Data Points
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retail
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
SINGLE INDUSTRY MODEL CUSTOM L AYERED MODEL
019
020
Comparing and Contrasting ModelsWant a straightforward look at how historical industry and custom models compare to custom layed models
Wersquove laid out the features and benefits of each in this chart
INDUSTRY MODELS CUSTOM MODELS CUSTOM LAYERED MODELS
Covers All Industries
Reflects Unique Business
Delivers Insights for a Combination of Offerings Channels amp Services
Can Be Adapted for an Individual Business without Changing the Model for All Clients Using That Model
Highly Accurate
Cost Effective
Time Efficient
Updated Regularly
Can Be Tested with Limited Data
020
021
21
FACTORING IN SENTIMENT
Getting to the WhyWhen humans look to tell effective storiesmdashwhether in film or in conversationmdashsentiment is
absolutely essential Itrsquos no different when your customer or employee is trying to tell you a
story about their experience via unstructured feedback Itrsquos your text analytics solutionrsquos job to
detect the emotion in their story identify it and help you understand what is causing that specific
sentiment so you can improve experiences But the specific method you use to help your solution
detect sentiment can affect the results you get and therefore your ability to act and transform
022
Rules-Based SentimentTraditionally the way text analytics identified senti-
ment was through a rules-based approach As the name
suggests this method applied rigid rules to understand
unstructured feedback Additionally changing these
rules or updating them required manual intervention
And even with those manual updates accuracy never
really improves
Adaptive Sentiment EngineThe idea that inspired the adaptive sentiment engine
approach is that rigid rules do not lend themselves to best
understand human emotions Instead adaptive sentiment
is AI driven meaning that it learns and improves over time
giving you more accurate results This engine continually
recognizes new terms and phrases as well meaning that
it grows and adapts in lockstep with your brand Check out
the difference below
023
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive Positive Negative NegativeSentiment Score 92
Confidence 98Sentiment Score 52
Confidence 97Sentiment Score -49
Confidence 41Sentiment Score -79
Confidence 94
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
RULES -BASED MACHINE LEARNING
NegativeNeutral
024
Now that you know what it takes to make your text analytics
solution truly world-class what next Well itrsquos time to make
sure that your greater experience program is set up to actually
drive improvement for your organization
Text analytics are just one piece of the puzzle when it comes
to using your VoC VoE or other experience initiatives to
power business growth When you look at the bigger picture
it becomes clear that you need a solid strategy for success
At InMoment we leverage the Continuous Improvement
Framework a method that helps us guide our clients through
designing listening understanding transforming and realizing
their goals for their experience initiative
If you want to learn more about this framework
and how text analytics fit into it you can read
more here
Like what yoursquove seen so far
Here are a few other pieces you may like
How You Listen Matters Modernizing Your Methods amp Approach to Customer Feedback
How to Prove the Business Value of Your CX Program
Building a Business Case for Your CFO (or Anyone Else)
Measuring Doesnrsquot Make You Taller
Therersquos more where that came from on
wwwinmomentcomresources
DRIVING IMPROVEMENT
024
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C
014
Human TranslationHuman translation is the method of having each comment translated
individually as it is submitted by the customer This type of manual
translation requires a long expensive process
Instead of real-time results there is usually a wait of 24 to 48 hours (or
more) to get the translated results And because human translation is
limited to languages that a specific translator can work with outsourcing
more complex languages to other translators will also add additional
costsmdashand more time
LETrsquoS GO OVER THE DIFFERENT OPTIONS
AND THEIR PROS AND CONS
015
Native Language LibrariesA native language library is a text analytics model created
entirely in a language of choice so the system analyzes and
tags comments in that native language instead of translating
them to English tagging them and exporting them back to the
original language Creating a native language library involves a
native speaker building out the initial text analytics model and
then constantly refining it to ensure that it remains up to date
Creating language libraries is a slow and expensive process
usually costing hundreds of thousands of dollars and taking
months or years to build Large amounts of data must be
collected collated and categorized in the language of choice
before the libraries can be built In addition each of these
libraries must be built and tuned to individual industries to
ensure their accuracy The main problem with these native
language libraries therefore is the lack of scalability
016
Machine Translated Human LedMachine translation is exactly what it sounds like It uses
machinesmdashtrained and monitored by humansmdashto translate
text When translating machine translation services look for
patterns and trends in millions of documents to provide the
best translation
Additionally many machine translation services are now using
a neural machine translation engine This means that rather
than translating individual words it uses a neural network
capable of deep learning to look at whole sentences and
understand the greater context These translation services are
highly accurate actionable cost efficient and scalable which
makes them the best option for an international VoC program
017
17
A ldquoMODELrdquo TEXT ANALY TICS SOLUTION
Tailoring Intelligence for Your BusinessThe next element of a text analytics solution to explore is the models you can employ to label
and organize your unstructured data These models have truly come a long way when it comes to their ability to deliver intelligence that is specifically tailored to your business so you can identify
areas that require action and deliver direct value to your bottom line
Traditionally brands were faced with the choice between two specific types of text analytics models industry models and custom models The definition of these two historical approaches
is self explanatory but letrsquos take a deeper dive
018
Industry ModelsAn industry model is developed by experts to reflect the
characteristics of a specific industry These models are
often designed by a vendor to be employed for every client
that fits that particular industry and therefore it often
misses brand-specific insights Additionally because the
same model is used for multiple brands the model canrsquot
be adapted for one brand without being adapted for all
However these models are a much more time-and-cost
effective solution when compared to custom models
Custom ModelsA custom model is built specifically for a certain company
gaining individualized context on the moments that set
a brand apart from its competitors While they are highly
accurate these models are more expensive to create and
maintain so they arenrsquot updated regularly Additionally they
require a high number of comments to test
019
Custom Layered ModelsThere are pros and cons to both of these historical
approaches but there is a new more effective
approach on the scene that gives experience
programs the best of both worlds custom layered
models Custom layered models take the accuracy of
custom models and the efficiency of industry models
to provide tailored accurate and effective results at a
fraction of the cost
How InMomentrsquos Custom Layered Models leverage
over 50 industry models to craft a personalized
tailored solution for each brand And when your text
analytics are built to suit your specific needs the
intelligence you gain from them is multiplied Take a
look at the difference
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retailecommerce
pharmacycustom tags
3xMore Data Points
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retail
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
SINGLE INDUSTRY MODEL CUSTOM L AYERED MODEL
019
020
Comparing and Contrasting ModelsWant a straightforward look at how historical industry and custom models compare to custom layed models
Wersquove laid out the features and benefits of each in this chart
INDUSTRY MODELS CUSTOM MODELS CUSTOM LAYERED MODELS
Covers All Industries
Reflects Unique Business
Delivers Insights for a Combination of Offerings Channels amp Services
Can Be Adapted for an Individual Business without Changing the Model for All Clients Using That Model
Highly Accurate
Cost Effective
Time Efficient
Updated Regularly
Can Be Tested with Limited Data
020
021
21
FACTORING IN SENTIMENT
Getting to the WhyWhen humans look to tell effective storiesmdashwhether in film or in conversationmdashsentiment is
absolutely essential Itrsquos no different when your customer or employee is trying to tell you a
story about their experience via unstructured feedback Itrsquos your text analytics solutionrsquos job to
detect the emotion in their story identify it and help you understand what is causing that specific
sentiment so you can improve experiences But the specific method you use to help your solution
detect sentiment can affect the results you get and therefore your ability to act and transform
022
Rules-Based SentimentTraditionally the way text analytics identified senti-
ment was through a rules-based approach As the name
suggests this method applied rigid rules to understand
unstructured feedback Additionally changing these
rules or updating them required manual intervention
And even with those manual updates accuracy never
really improves
Adaptive Sentiment EngineThe idea that inspired the adaptive sentiment engine
approach is that rigid rules do not lend themselves to best
understand human emotions Instead adaptive sentiment
is AI driven meaning that it learns and improves over time
giving you more accurate results This engine continually
recognizes new terms and phrases as well meaning that
it grows and adapts in lockstep with your brand Check out
the difference below
023
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive Positive Negative NegativeSentiment Score 92
Confidence 98Sentiment Score 52
Confidence 97Sentiment Score -49
Confidence 41Sentiment Score -79
Confidence 94
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
RULES -BASED MACHINE LEARNING
NegativeNeutral
024
Now that you know what it takes to make your text analytics
solution truly world-class what next Well itrsquos time to make
sure that your greater experience program is set up to actually
drive improvement for your organization
Text analytics are just one piece of the puzzle when it comes
to using your VoC VoE or other experience initiatives to
power business growth When you look at the bigger picture
it becomes clear that you need a solid strategy for success
At InMoment we leverage the Continuous Improvement
Framework a method that helps us guide our clients through
designing listening understanding transforming and realizing
their goals for their experience initiative
If you want to learn more about this framework
and how text analytics fit into it you can read
more here
Like what yoursquove seen so far
Here are a few other pieces you may like
How You Listen Matters Modernizing Your Methods amp Approach to Customer Feedback
How to Prove the Business Value of Your CX Program
Building a Business Case for Your CFO (or Anyone Else)
Measuring Doesnrsquot Make You Taller
Therersquos more where that came from on
wwwinmomentcomresources
DRIVING IMPROVEMENT
024
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C
015
Native Language LibrariesA native language library is a text analytics model created
entirely in a language of choice so the system analyzes and
tags comments in that native language instead of translating
them to English tagging them and exporting them back to the
original language Creating a native language library involves a
native speaker building out the initial text analytics model and
then constantly refining it to ensure that it remains up to date
Creating language libraries is a slow and expensive process
usually costing hundreds of thousands of dollars and taking
months or years to build Large amounts of data must be
collected collated and categorized in the language of choice
before the libraries can be built In addition each of these
libraries must be built and tuned to individual industries to
ensure their accuracy The main problem with these native
language libraries therefore is the lack of scalability
016
Machine Translated Human LedMachine translation is exactly what it sounds like It uses
machinesmdashtrained and monitored by humansmdashto translate
text When translating machine translation services look for
patterns and trends in millions of documents to provide the
best translation
Additionally many machine translation services are now using
a neural machine translation engine This means that rather
than translating individual words it uses a neural network
capable of deep learning to look at whole sentences and
understand the greater context These translation services are
highly accurate actionable cost efficient and scalable which
makes them the best option for an international VoC program
017
17
A ldquoMODELrdquo TEXT ANALY TICS SOLUTION
Tailoring Intelligence for Your BusinessThe next element of a text analytics solution to explore is the models you can employ to label
and organize your unstructured data These models have truly come a long way when it comes to their ability to deliver intelligence that is specifically tailored to your business so you can identify
areas that require action and deliver direct value to your bottom line
Traditionally brands were faced with the choice between two specific types of text analytics models industry models and custom models The definition of these two historical approaches
is self explanatory but letrsquos take a deeper dive
018
Industry ModelsAn industry model is developed by experts to reflect the
characteristics of a specific industry These models are
often designed by a vendor to be employed for every client
that fits that particular industry and therefore it often
misses brand-specific insights Additionally because the
same model is used for multiple brands the model canrsquot
be adapted for one brand without being adapted for all
However these models are a much more time-and-cost
effective solution when compared to custom models
Custom ModelsA custom model is built specifically for a certain company
gaining individualized context on the moments that set
a brand apart from its competitors While they are highly
accurate these models are more expensive to create and
maintain so they arenrsquot updated regularly Additionally they
require a high number of comments to test
019
Custom Layered ModelsThere are pros and cons to both of these historical
approaches but there is a new more effective
approach on the scene that gives experience
programs the best of both worlds custom layered
models Custom layered models take the accuracy of
custom models and the efficiency of industry models
to provide tailored accurate and effective results at a
fraction of the cost
How InMomentrsquos Custom Layered Models leverage
over 50 industry models to craft a personalized
tailored solution for each brand And when your text
analytics are built to suit your specific needs the
intelligence you gain from them is multiplied Take a
look at the difference
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retailecommerce
pharmacycustom tags
3xMore Data Points
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retail
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
SINGLE INDUSTRY MODEL CUSTOM L AYERED MODEL
019
020
Comparing and Contrasting ModelsWant a straightforward look at how historical industry and custom models compare to custom layed models
Wersquove laid out the features and benefits of each in this chart
INDUSTRY MODELS CUSTOM MODELS CUSTOM LAYERED MODELS
Covers All Industries
Reflects Unique Business
Delivers Insights for a Combination of Offerings Channels amp Services
Can Be Adapted for an Individual Business without Changing the Model for All Clients Using That Model
Highly Accurate
Cost Effective
Time Efficient
Updated Regularly
Can Be Tested with Limited Data
020
021
21
FACTORING IN SENTIMENT
Getting to the WhyWhen humans look to tell effective storiesmdashwhether in film or in conversationmdashsentiment is
absolutely essential Itrsquos no different when your customer or employee is trying to tell you a
story about their experience via unstructured feedback Itrsquos your text analytics solutionrsquos job to
detect the emotion in their story identify it and help you understand what is causing that specific
sentiment so you can improve experiences But the specific method you use to help your solution
detect sentiment can affect the results you get and therefore your ability to act and transform
022
Rules-Based SentimentTraditionally the way text analytics identified senti-
ment was through a rules-based approach As the name
suggests this method applied rigid rules to understand
unstructured feedback Additionally changing these
rules or updating them required manual intervention
And even with those manual updates accuracy never
really improves
Adaptive Sentiment EngineThe idea that inspired the adaptive sentiment engine
approach is that rigid rules do not lend themselves to best
understand human emotions Instead adaptive sentiment
is AI driven meaning that it learns and improves over time
giving you more accurate results This engine continually
recognizes new terms and phrases as well meaning that
it grows and adapts in lockstep with your brand Check out
the difference below
023
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive Positive Negative NegativeSentiment Score 92
Confidence 98Sentiment Score 52
Confidence 97Sentiment Score -49
Confidence 41Sentiment Score -79
Confidence 94
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
RULES -BASED MACHINE LEARNING
NegativeNeutral
024
Now that you know what it takes to make your text analytics
solution truly world-class what next Well itrsquos time to make
sure that your greater experience program is set up to actually
drive improvement for your organization
Text analytics are just one piece of the puzzle when it comes
to using your VoC VoE or other experience initiatives to
power business growth When you look at the bigger picture
it becomes clear that you need a solid strategy for success
At InMoment we leverage the Continuous Improvement
Framework a method that helps us guide our clients through
designing listening understanding transforming and realizing
their goals for their experience initiative
If you want to learn more about this framework
and how text analytics fit into it you can read
more here
Like what yoursquove seen so far
Here are a few other pieces you may like
How You Listen Matters Modernizing Your Methods amp Approach to Customer Feedback
How to Prove the Business Value of Your CX Program
Building a Business Case for Your CFO (or Anyone Else)
Measuring Doesnrsquot Make You Taller
Therersquos more where that came from on
wwwinmomentcomresources
DRIVING IMPROVEMENT
024
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C
016
Machine Translated Human LedMachine translation is exactly what it sounds like It uses
machinesmdashtrained and monitored by humansmdashto translate
text When translating machine translation services look for
patterns and trends in millions of documents to provide the
best translation
Additionally many machine translation services are now using
a neural machine translation engine This means that rather
than translating individual words it uses a neural network
capable of deep learning to look at whole sentences and
understand the greater context These translation services are
highly accurate actionable cost efficient and scalable which
makes them the best option for an international VoC program
017
17
A ldquoMODELrdquo TEXT ANALY TICS SOLUTION
Tailoring Intelligence for Your BusinessThe next element of a text analytics solution to explore is the models you can employ to label
and organize your unstructured data These models have truly come a long way when it comes to their ability to deliver intelligence that is specifically tailored to your business so you can identify
areas that require action and deliver direct value to your bottom line
Traditionally brands were faced with the choice between two specific types of text analytics models industry models and custom models The definition of these two historical approaches
is self explanatory but letrsquos take a deeper dive
018
Industry ModelsAn industry model is developed by experts to reflect the
characteristics of a specific industry These models are
often designed by a vendor to be employed for every client
that fits that particular industry and therefore it often
misses brand-specific insights Additionally because the
same model is used for multiple brands the model canrsquot
be adapted for one brand without being adapted for all
However these models are a much more time-and-cost
effective solution when compared to custom models
Custom ModelsA custom model is built specifically for a certain company
gaining individualized context on the moments that set
a brand apart from its competitors While they are highly
accurate these models are more expensive to create and
maintain so they arenrsquot updated regularly Additionally they
require a high number of comments to test
019
Custom Layered ModelsThere are pros and cons to both of these historical
approaches but there is a new more effective
approach on the scene that gives experience
programs the best of both worlds custom layered
models Custom layered models take the accuracy of
custom models and the efficiency of industry models
to provide tailored accurate and effective results at a
fraction of the cost
How InMomentrsquos Custom Layered Models leverage
over 50 industry models to craft a personalized
tailored solution for each brand And when your text
analytics are built to suit your specific needs the
intelligence you gain from them is multiplied Take a
look at the difference
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retailecommerce
pharmacycustom tags
3xMore Data Points
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retail
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
SINGLE INDUSTRY MODEL CUSTOM L AYERED MODEL
019
020
Comparing and Contrasting ModelsWant a straightforward look at how historical industry and custom models compare to custom layed models
Wersquove laid out the features and benefits of each in this chart
INDUSTRY MODELS CUSTOM MODELS CUSTOM LAYERED MODELS
Covers All Industries
Reflects Unique Business
Delivers Insights for a Combination of Offerings Channels amp Services
Can Be Adapted for an Individual Business without Changing the Model for All Clients Using That Model
Highly Accurate
Cost Effective
Time Efficient
Updated Regularly
Can Be Tested with Limited Data
020
021
21
FACTORING IN SENTIMENT
Getting to the WhyWhen humans look to tell effective storiesmdashwhether in film or in conversationmdashsentiment is
absolutely essential Itrsquos no different when your customer or employee is trying to tell you a
story about their experience via unstructured feedback Itrsquos your text analytics solutionrsquos job to
detect the emotion in their story identify it and help you understand what is causing that specific
sentiment so you can improve experiences But the specific method you use to help your solution
detect sentiment can affect the results you get and therefore your ability to act and transform
022
Rules-Based SentimentTraditionally the way text analytics identified senti-
ment was through a rules-based approach As the name
suggests this method applied rigid rules to understand
unstructured feedback Additionally changing these
rules or updating them required manual intervention
And even with those manual updates accuracy never
really improves
Adaptive Sentiment EngineThe idea that inspired the adaptive sentiment engine
approach is that rigid rules do not lend themselves to best
understand human emotions Instead adaptive sentiment
is AI driven meaning that it learns and improves over time
giving you more accurate results This engine continually
recognizes new terms and phrases as well meaning that
it grows and adapts in lockstep with your brand Check out
the difference below
023
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive Positive Negative NegativeSentiment Score 92
Confidence 98Sentiment Score 52
Confidence 97Sentiment Score -49
Confidence 41Sentiment Score -79
Confidence 94
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
RULES -BASED MACHINE LEARNING
NegativeNeutral
024
Now that you know what it takes to make your text analytics
solution truly world-class what next Well itrsquos time to make
sure that your greater experience program is set up to actually
drive improvement for your organization
Text analytics are just one piece of the puzzle when it comes
to using your VoC VoE or other experience initiatives to
power business growth When you look at the bigger picture
it becomes clear that you need a solid strategy for success
At InMoment we leverage the Continuous Improvement
Framework a method that helps us guide our clients through
designing listening understanding transforming and realizing
their goals for their experience initiative
If you want to learn more about this framework
and how text analytics fit into it you can read
more here
Like what yoursquove seen so far
Here are a few other pieces you may like
How You Listen Matters Modernizing Your Methods amp Approach to Customer Feedback
How to Prove the Business Value of Your CX Program
Building a Business Case for Your CFO (or Anyone Else)
Measuring Doesnrsquot Make You Taller
Therersquos more where that came from on
wwwinmomentcomresources
DRIVING IMPROVEMENT
024
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C
017
17
A ldquoMODELrdquo TEXT ANALY TICS SOLUTION
Tailoring Intelligence for Your BusinessThe next element of a text analytics solution to explore is the models you can employ to label
and organize your unstructured data These models have truly come a long way when it comes to their ability to deliver intelligence that is specifically tailored to your business so you can identify
areas that require action and deliver direct value to your bottom line
Traditionally brands were faced with the choice between two specific types of text analytics models industry models and custom models The definition of these two historical approaches
is self explanatory but letrsquos take a deeper dive
018
Industry ModelsAn industry model is developed by experts to reflect the
characteristics of a specific industry These models are
often designed by a vendor to be employed for every client
that fits that particular industry and therefore it often
misses brand-specific insights Additionally because the
same model is used for multiple brands the model canrsquot
be adapted for one brand without being adapted for all
However these models are a much more time-and-cost
effective solution when compared to custom models
Custom ModelsA custom model is built specifically for a certain company
gaining individualized context on the moments that set
a brand apart from its competitors While they are highly
accurate these models are more expensive to create and
maintain so they arenrsquot updated regularly Additionally they
require a high number of comments to test
019
Custom Layered ModelsThere are pros and cons to both of these historical
approaches but there is a new more effective
approach on the scene that gives experience
programs the best of both worlds custom layered
models Custom layered models take the accuracy of
custom models and the efficiency of industry models
to provide tailored accurate and effective results at a
fraction of the cost
How InMomentrsquos Custom Layered Models leverage
over 50 industry models to craft a personalized
tailored solution for each brand And when your text
analytics are built to suit your specific needs the
intelligence you gain from them is multiplied Take a
look at the difference
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retailecommerce
pharmacycustom tags
3xMore Data Points
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retail
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
SINGLE INDUSTRY MODEL CUSTOM L AYERED MODEL
019
020
Comparing and Contrasting ModelsWant a straightforward look at how historical industry and custom models compare to custom layed models
Wersquove laid out the features and benefits of each in this chart
INDUSTRY MODELS CUSTOM MODELS CUSTOM LAYERED MODELS
Covers All Industries
Reflects Unique Business
Delivers Insights for a Combination of Offerings Channels amp Services
Can Be Adapted for an Individual Business without Changing the Model for All Clients Using That Model
Highly Accurate
Cost Effective
Time Efficient
Updated Regularly
Can Be Tested with Limited Data
020
021
21
FACTORING IN SENTIMENT
Getting to the WhyWhen humans look to tell effective storiesmdashwhether in film or in conversationmdashsentiment is
absolutely essential Itrsquos no different when your customer or employee is trying to tell you a
story about their experience via unstructured feedback Itrsquos your text analytics solutionrsquos job to
detect the emotion in their story identify it and help you understand what is causing that specific
sentiment so you can improve experiences But the specific method you use to help your solution
detect sentiment can affect the results you get and therefore your ability to act and transform
022
Rules-Based SentimentTraditionally the way text analytics identified senti-
ment was through a rules-based approach As the name
suggests this method applied rigid rules to understand
unstructured feedback Additionally changing these
rules or updating them required manual intervention
And even with those manual updates accuracy never
really improves
Adaptive Sentiment EngineThe idea that inspired the adaptive sentiment engine
approach is that rigid rules do not lend themselves to best
understand human emotions Instead adaptive sentiment
is AI driven meaning that it learns and improves over time
giving you more accurate results This engine continually
recognizes new terms and phrases as well meaning that
it grows and adapts in lockstep with your brand Check out
the difference below
023
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive Positive Negative NegativeSentiment Score 92
Confidence 98Sentiment Score 52
Confidence 97Sentiment Score -49
Confidence 41Sentiment Score -79
Confidence 94
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
RULES -BASED MACHINE LEARNING
NegativeNeutral
024
Now that you know what it takes to make your text analytics
solution truly world-class what next Well itrsquos time to make
sure that your greater experience program is set up to actually
drive improvement for your organization
Text analytics are just one piece of the puzzle when it comes
to using your VoC VoE or other experience initiatives to
power business growth When you look at the bigger picture
it becomes clear that you need a solid strategy for success
At InMoment we leverage the Continuous Improvement
Framework a method that helps us guide our clients through
designing listening understanding transforming and realizing
their goals for their experience initiative
If you want to learn more about this framework
and how text analytics fit into it you can read
more here
Like what yoursquove seen so far
Here are a few other pieces you may like
How You Listen Matters Modernizing Your Methods amp Approach to Customer Feedback
How to Prove the Business Value of Your CX Program
Building a Business Case for Your CFO (or Anyone Else)
Measuring Doesnrsquot Make You Taller
Therersquos more where that came from on
wwwinmomentcomresources
DRIVING IMPROVEMENT
024
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C
018
Industry ModelsAn industry model is developed by experts to reflect the
characteristics of a specific industry These models are
often designed by a vendor to be employed for every client
that fits that particular industry and therefore it often
misses brand-specific insights Additionally because the
same model is used for multiple brands the model canrsquot
be adapted for one brand without being adapted for all
However these models are a much more time-and-cost
effective solution when compared to custom models
Custom ModelsA custom model is built specifically for a certain company
gaining individualized context on the moments that set
a brand apart from its competitors While they are highly
accurate these models are more expensive to create and
maintain so they arenrsquot updated regularly Additionally they
require a high number of comments to test
019
Custom Layered ModelsThere are pros and cons to both of these historical
approaches but there is a new more effective
approach on the scene that gives experience
programs the best of both worlds custom layered
models Custom layered models take the accuracy of
custom models and the efficiency of industry models
to provide tailored accurate and effective results at a
fraction of the cost
How InMomentrsquos Custom Layered Models leverage
over 50 industry models to craft a personalized
tailored solution for each brand And when your text
analytics are built to suit your specific needs the
intelligence you gain from them is multiplied Take a
look at the difference
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retailecommerce
pharmacycustom tags
3xMore Data Points
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retail
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
SINGLE INDUSTRY MODEL CUSTOM L AYERED MODEL
019
020
Comparing and Contrasting ModelsWant a straightforward look at how historical industry and custom models compare to custom layed models
Wersquove laid out the features and benefits of each in this chart
INDUSTRY MODELS CUSTOM MODELS CUSTOM LAYERED MODELS
Covers All Industries
Reflects Unique Business
Delivers Insights for a Combination of Offerings Channels amp Services
Can Be Adapted for an Individual Business without Changing the Model for All Clients Using That Model
Highly Accurate
Cost Effective
Time Efficient
Updated Regularly
Can Be Tested with Limited Data
020
021
21
FACTORING IN SENTIMENT
Getting to the WhyWhen humans look to tell effective storiesmdashwhether in film or in conversationmdashsentiment is
absolutely essential Itrsquos no different when your customer or employee is trying to tell you a
story about their experience via unstructured feedback Itrsquos your text analytics solutionrsquos job to
detect the emotion in their story identify it and help you understand what is causing that specific
sentiment so you can improve experiences But the specific method you use to help your solution
detect sentiment can affect the results you get and therefore your ability to act and transform
022
Rules-Based SentimentTraditionally the way text analytics identified senti-
ment was through a rules-based approach As the name
suggests this method applied rigid rules to understand
unstructured feedback Additionally changing these
rules or updating them required manual intervention
And even with those manual updates accuracy never
really improves
Adaptive Sentiment EngineThe idea that inspired the adaptive sentiment engine
approach is that rigid rules do not lend themselves to best
understand human emotions Instead adaptive sentiment
is AI driven meaning that it learns and improves over time
giving you more accurate results This engine continually
recognizes new terms and phrases as well meaning that
it grows and adapts in lockstep with your brand Check out
the difference below
023
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive Positive Negative NegativeSentiment Score 92
Confidence 98Sentiment Score 52
Confidence 97Sentiment Score -49
Confidence 41Sentiment Score -79
Confidence 94
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
RULES -BASED MACHINE LEARNING
NegativeNeutral
024
Now that you know what it takes to make your text analytics
solution truly world-class what next Well itrsquos time to make
sure that your greater experience program is set up to actually
drive improvement for your organization
Text analytics are just one piece of the puzzle when it comes
to using your VoC VoE or other experience initiatives to
power business growth When you look at the bigger picture
it becomes clear that you need a solid strategy for success
At InMoment we leverage the Continuous Improvement
Framework a method that helps us guide our clients through
designing listening understanding transforming and realizing
their goals for their experience initiative
If you want to learn more about this framework
and how text analytics fit into it you can read
more here
Like what yoursquove seen so far
Here are a few other pieces you may like
How You Listen Matters Modernizing Your Methods amp Approach to Customer Feedback
How to Prove the Business Value of Your CX Program
Building a Business Case for Your CFO (or Anyone Else)
Measuring Doesnrsquot Make You Taller
Therersquos more where that came from on
wwwinmomentcomresources
DRIVING IMPROVEMENT
024
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C
019
Custom Layered ModelsThere are pros and cons to both of these historical
approaches but there is a new more effective
approach on the scene that gives experience
programs the best of both worlds custom layered
models Custom layered models take the accuracy of
custom models and the efficiency of industry models
to provide tailored accurate and effective results at a
fraction of the cost
How InMomentrsquos Custom Layered Models leverage
over 50 industry models to craft a personalized
tailored solution for each brand And when your text
analytics are built to suit your specific needs the
intelligence you gain from them is multiplied Take a
look at the difference
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retailecommerce
pharmacycustom tags
3xMore Data Points
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
Please tell us more about your most recent experience
DETAIL STRENGTH
INDUSTRY
retail
I refilled my prescription through the Health
Rewards mobile app which was simple but I
didnrsquot get a notification when it was ready
When I arrived I picked up a few extra things
I needed in the grocery section because
there was a flash sale Unfortunately I had to
wait in line forever to talk to the pharmacist
even though the store was quiet
SINGLE INDUSTRY MODEL CUSTOM L AYERED MODEL
019
020
Comparing and Contrasting ModelsWant a straightforward look at how historical industry and custom models compare to custom layed models
Wersquove laid out the features and benefits of each in this chart
INDUSTRY MODELS CUSTOM MODELS CUSTOM LAYERED MODELS
Covers All Industries
Reflects Unique Business
Delivers Insights for a Combination of Offerings Channels amp Services
Can Be Adapted for an Individual Business without Changing the Model for All Clients Using That Model
Highly Accurate
Cost Effective
Time Efficient
Updated Regularly
Can Be Tested with Limited Data
020
021
21
FACTORING IN SENTIMENT
Getting to the WhyWhen humans look to tell effective storiesmdashwhether in film or in conversationmdashsentiment is
absolutely essential Itrsquos no different when your customer or employee is trying to tell you a
story about their experience via unstructured feedback Itrsquos your text analytics solutionrsquos job to
detect the emotion in their story identify it and help you understand what is causing that specific
sentiment so you can improve experiences But the specific method you use to help your solution
detect sentiment can affect the results you get and therefore your ability to act and transform
022
Rules-Based SentimentTraditionally the way text analytics identified senti-
ment was through a rules-based approach As the name
suggests this method applied rigid rules to understand
unstructured feedback Additionally changing these
rules or updating them required manual intervention
And even with those manual updates accuracy never
really improves
Adaptive Sentiment EngineThe idea that inspired the adaptive sentiment engine
approach is that rigid rules do not lend themselves to best
understand human emotions Instead adaptive sentiment
is AI driven meaning that it learns and improves over time
giving you more accurate results This engine continually
recognizes new terms and phrases as well meaning that
it grows and adapts in lockstep with your brand Check out
the difference below
023
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive Positive Negative NegativeSentiment Score 92
Confidence 98Sentiment Score 52
Confidence 97Sentiment Score -49
Confidence 41Sentiment Score -79
Confidence 94
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
RULES -BASED MACHINE LEARNING
NegativeNeutral
024
Now that you know what it takes to make your text analytics
solution truly world-class what next Well itrsquos time to make
sure that your greater experience program is set up to actually
drive improvement for your organization
Text analytics are just one piece of the puzzle when it comes
to using your VoC VoE or other experience initiatives to
power business growth When you look at the bigger picture
it becomes clear that you need a solid strategy for success
At InMoment we leverage the Continuous Improvement
Framework a method that helps us guide our clients through
designing listening understanding transforming and realizing
their goals for their experience initiative
If you want to learn more about this framework
and how text analytics fit into it you can read
more here
Like what yoursquove seen so far
Here are a few other pieces you may like
How You Listen Matters Modernizing Your Methods amp Approach to Customer Feedback
How to Prove the Business Value of Your CX Program
Building a Business Case for Your CFO (or Anyone Else)
Measuring Doesnrsquot Make You Taller
Therersquos more where that came from on
wwwinmomentcomresources
DRIVING IMPROVEMENT
024
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C
020
Comparing and Contrasting ModelsWant a straightforward look at how historical industry and custom models compare to custom layed models
Wersquove laid out the features and benefits of each in this chart
INDUSTRY MODELS CUSTOM MODELS CUSTOM LAYERED MODELS
Covers All Industries
Reflects Unique Business
Delivers Insights for a Combination of Offerings Channels amp Services
Can Be Adapted for an Individual Business without Changing the Model for All Clients Using That Model
Highly Accurate
Cost Effective
Time Efficient
Updated Regularly
Can Be Tested with Limited Data
020
021
21
FACTORING IN SENTIMENT
Getting to the WhyWhen humans look to tell effective storiesmdashwhether in film or in conversationmdashsentiment is
absolutely essential Itrsquos no different when your customer or employee is trying to tell you a
story about their experience via unstructured feedback Itrsquos your text analytics solutionrsquos job to
detect the emotion in their story identify it and help you understand what is causing that specific
sentiment so you can improve experiences But the specific method you use to help your solution
detect sentiment can affect the results you get and therefore your ability to act and transform
022
Rules-Based SentimentTraditionally the way text analytics identified senti-
ment was through a rules-based approach As the name
suggests this method applied rigid rules to understand
unstructured feedback Additionally changing these
rules or updating them required manual intervention
And even with those manual updates accuracy never
really improves
Adaptive Sentiment EngineThe idea that inspired the adaptive sentiment engine
approach is that rigid rules do not lend themselves to best
understand human emotions Instead adaptive sentiment
is AI driven meaning that it learns and improves over time
giving you more accurate results This engine continually
recognizes new terms and phrases as well meaning that
it grows and adapts in lockstep with your brand Check out
the difference below
023
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive Positive Negative NegativeSentiment Score 92
Confidence 98Sentiment Score 52
Confidence 97Sentiment Score -49
Confidence 41Sentiment Score -79
Confidence 94
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
RULES -BASED MACHINE LEARNING
NegativeNeutral
024
Now that you know what it takes to make your text analytics
solution truly world-class what next Well itrsquos time to make
sure that your greater experience program is set up to actually
drive improvement for your organization
Text analytics are just one piece of the puzzle when it comes
to using your VoC VoE or other experience initiatives to
power business growth When you look at the bigger picture
it becomes clear that you need a solid strategy for success
At InMoment we leverage the Continuous Improvement
Framework a method that helps us guide our clients through
designing listening understanding transforming and realizing
their goals for their experience initiative
If you want to learn more about this framework
and how text analytics fit into it you can read
more here
Like what yoursquove seen so far
Here are a few other pieces you may like
How You Listen Matters Modernizing Your Methods amp Approach to Customer Feedback
How to Prove the Business Value of Your CX Program
Building a Business Case for Your CFO (or Anyone Else)
Measuring Doesnrsquot Make You Taller
Therersquos more where that came from on
wwwinmomentcomresources
DRIVING IMPROVEMENT
024
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C
021
21
FACTORING IN SENTIMENT
Getting to the WhyWhen humans look to tell effective storiesmdashwhether in film or in conversationmdashsentiment is
absolutely essential Itrsquos no different when your customer or employee is trying to tell you a
story about their experience via unstructured feedback Itrsquos your text analytics solutionrsquos job to
detect the emotion in their story identify it and help you understand what is causing that specific
sentiment so you can improve experiences But the specific method you use to help your solution
detect sentiment can affect the results you get and therefore your ability to act and transform
022
Rules-Based SentimentTraditionally the way text analytics identified senti-
ment was through a rules-based approach As the name
suggests this method applied rigid rules to understand
unstructured feedback Additionally changing these
rules or updating them required manual intervention
And even with those manual updates accuracy never
really improves
Adaptive Sentiment EngineThe idea that inspired the adaptive sentiment engine
approach is that rigid rules do not lend themselves to best
understand human emotions Instead adaptive sentiment
is AI driven meaning that it learns and improves over time
giving you more accurate results This engine continually
recognizes new terms and phrases as well meaning that
it grows and adapts in lockstep with your brand Check out
the difference below
023
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive Positive Negative NegativeSentiment Score 92
Confidence 98Sentiment Score 52
Confidence 97Sentiment Score -49
Confidence 41Sentiment Score -79
Confidence 94
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
RULES -BASED MACHINE LEARNING
NegativeNeutral
024
Now that you know what it takes to make your text analytics
solution truly world-class what next Well itrsquos time to make
sure that your greater experience program is set up to actually
drive improvement for your organization
Text analytics are just one piece of the puzzle when it comes
to using your VoC VoE or other experience initiatives to
power business growth When you look at the bigger picture
it becomes clear that you need a solid strategy for success
At InMoment we leverage the Continuous Improvement
Framework a method that helps us guide our clients through
designing listening understanding transforming and realizing
their goals for their experience initiative
If you want to learn more about this framework
and how text analytics fit into it you can read
more here
Like what yoursquove seen so far
Here are a few other pieces you may like
How You Listen Matters Modernizing Your Methods amp Approach to Customer Feedback
How to Prove the Business Value of Your CX Program
Building a Business Case for Your CFO (or Anyone Else)
Measuring Doesnrsquot Make You Taller
Therersquos more where that came from on
wwwinmomentcomresources
DRIVING IMPROVEMENT
024
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C
022
Rules-Based SentimentTraditionally the way text analytics identified senti-
ment was through a rules-based approach As the name
suggests this method applied rigid rules to understand
unstructured feedback Additionally changing these
rules or updating them required manual intervention
And even with those manual updates accuracy never
really improves
Adaptive Sentiment EngineThe idea that inspired the adaptive sentiment engine
approach is that rigid rules do not lend themselves to best
understand human emotions Instead adaptive sentiment
is AI driven meaning that it learns and improves over time
giving you more accurate results This engine continually
recognizes new terms and phrases as well meaning that
it grows and adapts in lockstep with your brand Check out
the difference below
023
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive Positive Negative NegativeSentiment Score 92
Confidence 98Sentiment Score 52
Confidence 97Sentiment Score -49
Confidence 41Sentiment Score -79
Confidence 94
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
RULES -BASED MACHINE LEARNING
NegativeNeutral
024
Now that you know what it takes to make your text analytics
solution truly world-class what next Well itrsquos time to make
sure that your greater experience program is set up to actually
drive improvement for your organization
Text analytics are just one piece of the puzzle when it comes
to using your VoC VoE or other experience initiatives to
power business growth When you look at the bigger picture
it becomes clear that you need a solid strategy for success
At InMoment we leverage the Continuous Improvement
Framework a method that helps us guide our clients through
designing listening understanding transforming and realizing
their goals for their experience initiative
If you want to learn more about this framework
and how text analytics fit into it you can read
more here
Like what yoursquove seen so far
Here are a few other pieces you may like
How You Listen Matters Modernizing Your Methods amp Approach to Customer Feedback
How to Prove the Business Value of Your CX Program
Building a Business Case for Your CFO (or Anyone Else)
Measuring Doesnrsquot Make You Taller
Therersquos more where that came from on
wwwinmomentcomresources
DRIVING IMPROVEMENT
024
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C
023
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
Please tell us more about your most recent experience
DETAIL STRENGTH
PHRASE SENTIMENT
Positive Positive Negative NegativeSentiment Score 92
Confidence 98Sentiment Score 52
Confidence 97Sentiment Score -49
Confidence 41Sentiment Score -79
Confidence 94
I unloaded my bike and was able to leave
my trailer on the car while I did other
shopping but there are a few things
that need attention in my opinion For
example no one approached me to help
me I had to find an employee myself
But she gave me prices and options
without wasting my time
RULES -BASED MACHINE LEARNING
NegativeNeutral
024
Now that you know what it takes to make your text analytics
solution truly world-class what next Well itrsquos time to make
sure that your greater experience program is set up to actually
drive improvement for your organization
Text analytics are just one piece of the puzzle when it comes
to using your VoC VoE or other experience initiatives to
power business growth When you look at the bigger picture
it becomes clear that you need a solid strategy for success
At InMoment we leverage the Continuous Improvement
Framework a method that helps us guide our clients through
designing listening understanding transforming and realizing
their goals for their experience initiative
If you want to learn more about this framework
and how text analytics fit into it you can read
more here
Like what yoursquove seen so far
Here are a few other pieces you may like
How You Listen Matters Modernizing Your Methods amp Approach to Customer Feedback
How to Prove the Business Value of Your CX Program
Building a Business Case for Your CFO (or Anyone Else)
Measuring Doesnrsquot Make You Taller
Therersquos more where that came from on
wwwinmomentcomresources
DRIVING IMPROVEMENT
024
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C
024
Now that you know what it takes to make your text analytics
solution truly world-class what next Well itrsquos time to make
sure that your greater experience program is set up to actually
drive improvement for your organization
Text analytics are just one piece of the puzzle when it comes
to using your VoC VoE or other experience initiatives to
power business growth When you look at the bigger picture
it becomes clear that you need a solid strategy for success
At InMoment we leverage the Continuous Improvement
Framework a method that helps us guide our clients through
designing listening understanding transforming and realizing
their goals for their experience initiative
If you want to learn more about this framework
and how text analytics fit into it you can read
more here
Like what yoursquove seen so far
Here are a few other pieces you may like
How You Listen Matters Modernizing Your Methods amp Approach to Customer Feedback
How to Prove the Business Value of Your CX Program
Building a Business Case for Your CFO (or Anyone Else)
Measuring Doesnrsquot Make You Taller
Therersquos more where that came from on
wwwinmomentcomresources
DRIVING IMPROVEMENT
024
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C
025
NORTH AMERICA1 800 530 4251
About InMoment
Improving experiences is why InMomentTM exists Our mission is to help our clients improve experiences at the intersection
of valuemdashwhere customer employee and business needs come together The heart of what we do is connect our clients
with what matters most through a unique combination of data technology and human expertise With our hyper-modern
technology platform decades of domain authority and global teams of experts we uniquely deliver a focus on Experience
Improvement (XI) to help our clients own the moments that matter Take a moment and learn more at inmomentcom
APAC+61 (2) 8397 8131
UK amp IRELAND+44 (0) 1494 590 600
GERMANY+49 (0) 40 369 833 0
D M - 0 1 0 6 6 - 0 2 | copy 2 0 2 0 I N M O M E N T I N C